- What the OCUP2 Advanced Exam Looks Like
- The Eleven Content Areas at a Glance
- Common Structure (21%)
- Classification (14%) and the MOF & Metamodeling (12%)
- Activities, Interactions and Actions (25% Combined)
- Structured Classifiers, StateMachines and Common Behavior
- Alf and fUML (12% Combined)
- Reading the Specifications Without Drowning
- Sequencing Your Preparation by Domain Weight
- Registration, Fees and Validity
- Frequently Asked Questions
- The exam has 11 published content areas; Common Structure alone carries 21% of the weighting.
- You need at least 57 correct answers out of 90 multiple-choice questions to pass.
- MOF & Metamodeling is a separate 12% area, larger than Activities, Interactions or Actions on their own.
- Alf and fUML together account for 12%, so skipping them costs as much as ignoring MOF.
What the OCUP2 Advanced Exam Looks Like
OMG Certified UML Professional 2: Advanced (OCUP2 Advanced) is the top level of the OMG UML certification program. It is aimed at practitioners who go beyond drawing diagrams and need to reason about what the UML specification actually says. Before you plan anything, internalize the format: 90 multiple-choice questions that use text and images, a passing threshold of at least 57 correct answers (listed as at least 63%), and a time allowance of 105 minutes in native English-speaking countries or 135 minutes elsewhere. The 135-minute allowance should be confirmed with the exam provider when you book.
The "images" part matters. Many questions show a diagram fragment and ask what it means, whether it is well-formed, or what behavior it implies. That is a different skill from recalling a definition. If you want a deeper view of how demanding this is in practice, see our analysis in How Hard Is the OCUP2 Exam? Complete Difficulty Guide 2026, and for the arithmetic behind the cut score read OCUP2 Passing Score 2026: Exactly What You Need to Pass.
The Eleven Content Areas at a Glance
The issuer publishes eleven content areas whose weights total 100%. The table below reproduces them exactly. Note that the distribution is lopsided: the top three areas make up 47% of the exam, while the bottom four make up only 20%.
| # | Content Area | Weight |
|---|---|---|
| 1 | Common Structure | 21% |
| 2 | Classification | 14% |
| 3 | The MOF & Metamodeling | 12% |
| 4 | Activities | 9% |
| 5 | Interactions | 9% |
| 6 | Structured Classifiers | 8% |
| 7 | Actions | 7% |
| 8 | Alf | 6% |
| 9 | fUML | 6% |
| 10 | StateMachines | 5% |
| 11 | Common Behavior | 3% |
One caution: the weights are the issuer's published allocation. Any per-area question counts you see in practice material are editorial estimates, not official figures, so treat them as a planning aid only. For the full picture of how this guide fits with the rest of our coverage, the OCUP2 Study Guide 2026: How to Pass on Your First Attempt walks through the end-to-end preparation process.
Common Structure (21%)
The heaviest area on the exam covers the building blocks that nearly every other part of UML depends on. Because it is foundational, questions here often probe precise semantics rather than notation: how elements are owned, how names and namespaces behave, how relationships are derived, and how constraints and multiplicities are interpreted.
Common Structure: what to master
Expect questions that test the metamodel-level meaning of shared constructs, not just their appearance on a diagram.
- Ownership, namespaces, and how elements are named and resolved
- Multiplicity, ordering, and uniqueness as they apply to properties and parameters
- Constraints and how they attach to the elements they restrict
- Packages, imports, merges, and visibility rules
- Dependencies and how their varieties differ in meaning
Since this single area is worth more than any two of the middle-weight behavioral areas combined, a weak showing here is very hard to recover from elsewhere. Candidates who rushed through it in the Foundation and Intermediate exams often find the Advanced questions expect noticeably more rigor.
Classification (14%) and the MOF & Metamodeling (12%)
Classification
Classification deals with how classifiers, their features, and their relationships are defined: generalization and substitutability, redefinition, specialization of properties and operations, and how instances relate to classifiers. Advanced questions tend to present a small model and ask which statement about it is true, or whether a particular redefinition or generalization set is valid. Slow, careful reading of the diagram is half the battle.
Classification: high-value points
- Generalization sets: disjoint versus overlapping, complete versus incomplete
- Redefinition versus simple subsetting, and what each permits
- Association ends, navigability, and ownership of ends
- Interfaces, realization, and how required and provided interfaces differ
- Instances, slots, and how they conform to their classifiers
The MOF & Metamodeling
This is the area that most separates Advanced from the earlier levels, and it is published as its own line item at 12%. It asks you to think about UML as a model of models: the layered metamodeling architecture, how UML itself is defined using MOF, and how extension mechanisms such as profiles, stereotypes, and tagged values fit into that architecture. Candidates who have only ever used UML as a drawing tool tend to find this area unfamiliar.
If you are weighing whether this level of depth is worth the effort for your career, our Is the OCUP2 Certification Worth It? Complete ROI Analysis 2026 lays out the trade-offs honestly.
Activities, Interactions and Actions (25% Combined)
Three behavioral areas together make up a quarter of the exam. They overlap conceptually, so studying them as a connected group pays off.
Activities (9%)
Activity questions center on token flow and execution semantics rather than diagram layout.
- How tokens move through nodes and edges, and when a node can fire
- Fork, join, decision, and merge behavior and their guard conditions
- Object nodes, pins, and buffering behavior
- Partitions, interruptible regions, and exception handling
Interactions (9%)
Interactions concern the exchange of messages between participants over time.
- Lifelines, messages, and the difference between synchronous and asynchronous communication
- Combined fragments and their operators
- Interaction uses, gates, and how interactions compose
- Reading traces: which sequences of events a diagram allows or forbids
Actions (7%)
Actions are the fine-grained executable units that Activities and other behaviors are built from, and they are the bridge to the Alf and fUML areas.
- Object, link, and variable actions and what each reads or writes
- Invocation actions, including how operation calls and signal sends differ
- Input and output pins and how values flow between actions
Because Actions underpin both Activities and the executable-model areas, studying them early makes Alf and fUML considerably easier later.
Structured Classifiers, StateMachines and Common Behavior
These three areas total 16% and are individually smaller, but they still reward focused review.
Structured Classifiers (8%)
This area covers internal structure: parts, ports, connectors, and collaborations. Typical questions ask how a connector links ports, what a port's provided and required interfaces imply, and how a classifier's internal structure delegates behavior. Pay particular attention to the distinction between a part and a reference, and to how connectors are typed.
StateMachines (5%)
StateMachines are a smaller slice than many candidates expect given how prominent state diagrams are in everyday modeling. Questions focus on execution semantics: how transitions are selected, how composite and orthogonal states behave, entry and exit ordering, and the role of pseudostates such as history, fork, join, and choice. Memorize the precise sequence in which exit, effect, and entry behaviors occur when a transition fires.
Common Behavior (3%)
The smallest area covers behavior concepts shared across the different behavior formalisms, including how behaviors relate to their context and how events and triggers are defined. It is only 3%, but the concepts it introduces recur in the Activities, Interactions, and StateMachines questions, so a quick pass early in your preparation helps everywhere else.
Key Takeaway
Do not mistake low weight for low value. StateMachines (5%) and Common Behavior (3%) are cheap to learn well, and together they add up to 8% of the exam. Neglecting them gives away points that are among the easiest to secure.
Alf and fUML (12% Combined)
These two areas are what make the Advanced exam distinctive. Alf is the Action Language for Foundational UML, a textual syntax for expressing executable behavior, and fUML is the Foundational UML subset with precise execution semantics. Each is weighted at 6%.
Alf (6%)
Expect to read short Alf snippets and determine what they do or which model construct they correspond to.
- Basic expression and statement syntax
- How Alf constructs map onto UML actions and activities
- Declarations of classes, operations, and behaviors in textual form
fUML (6%)
fUML questions concern the precise execution semantics of the foundational subset.
- Which UML constructs fall inside the foundational subset
- How activities execute under the fUML execution model
- Core semantic concepts such as how objects, values, and behaviors are handled at runtime
Because these are covered by the standard references (Alf v1.1 and fUML v1.5), your preparation here should lean on the specified portions of those specifications rather than on general modeling experience. People who have never worked with executable UML should budget extra time for this block.
Reading the Specifications Without Drowning
The standard references for the exam are UML v2.5.1, Alf v1.1 and fUML v1.5. These are large documents, and attempting to read them cover to cover is one of the most common ways candidates burn out. The issuer's sheet lists the specific chapters and subsections that are in scope, and that list, not the full text of each specification, defines the exam boundary. The same applies to recommended study-book chapters: not every chapter maps to something examinable.
For a compact refresher once you have worked through the specifications, the OCUP2 Cheat Sheet 2026: One-Page Review of Must-Know Facts is a useful final-week companion.
Sequencing Your Preparation by Domain Weight
Rather than a generic schedule, order your study by dependency and weight. The structural areas feed everything else, so they come first; the executable-model areas build on Actions, so they come last. Here is one reasonable arrangement.
Common Structure and Common Behavior
- Cover the 21% foundation area thoroughly
- Skim Common Behavior (3%) so later behavioral topics have context
Classification and MOF & Metamodeling
- Work through the 14% and 12% areas back to back, since both rely on metamodel thinking
- Spend extra time on profiles and the layered architecture
Actions, Activities, Interactions
- Start with Actions (7%) so Activities (9%) make sense
- Finish with Interactions (9%) and practice trace-reading
Structured Classifiers, StateMachines, Alf, fUML
- Cover the remaining structural and state areas
- Close with Alf and fUML once Actions are fresh
Adjust the pace to your background; someone already fluent in executable UML can compress the final week, while someone new to metamodeling may want to stretch Week 2. If you are looking for a broader plan, our OCUP2 training overview covers the preparation options available. After each block, test yourself with timed questions on the OCUP2 practice test site to confirm the material has stuck.
Registration, Fees and Validity
The practical details matter because they affect when and how you can sit the exam.
- Prerequisites: You must have passed the UML 2 Foundation and Intermediate examinations before attempting Advanced.
- Price: The listed fee is US$350 plus taxes, with regional currency equivalents. See OCUP2 Certification Cost 2026: Complete Pricing Breakdown for the full picture.
- Booking: Through Pearson VUE, either at a test center or online, subject to the provider's system test.
- Language: English is the published exam language, and translation apps are prohibited.
- Cancellation: Cancelling more than 24 hours before the scheduled exam permits a full refund.
- Retakes: A candidate who fails should check the score report for a 20% retake discount code.
- Validity: Certification expires three years after the examination; taking the same or a higher-level exam extends validity.
Scheduling specifics are covered in OCUP2 Exam Dates 2026: Testing Windows, Deadlines & Scheduling. Details such as retake waiting intervals and which reference materials, if any, are permitted during the exam are not confirmed in the issuer's sheet, so verify them with Pearson VUE before test day rather than relying on assumptions.
Frequently Asked Questions
There are eleven published content areas: Common Structure, Classification, The MOF & Metamodeling, Activities, Interactions, Structured Classifiers, Actions, Alf, fUML, StateMachines, and Common Behavior. Their weights total 100%, with Common Structure the largest at 21%.
You need at least 57 correct answers out of 90 questions, which the issuer lists as at least 63%. The exam is entirely multiple choice, using text and images. For the detailed breakdown, see our passing score guide.
Start with Common Structure, since it is the heaviest at 21% and underpins most other areas. Then move to Classification and MOF & Metamodeling before tackling the behavioral areas, and leave Alf and fUML for last because they build on Actions.
No. The issuer's sheet lists the specific chapters and subsections of UML v2.5.1, Alf v1.1 and fUML v1.5 that are examinable. Use that list as your boundary instead of reading each specification in full.
Certification expires three years after the examination. According to the issuer, taking the same or a higher-level exam extends validity. For context on what the credential can mean for your career, see OCUP2 jobs.