AWS Developer Associate Exam Format and Time Management

The current exam structure

The current Developer Associate guide describes a 130-minute exam with 65 presented questions. Fifty count toward the score and 15 are unscored pretest items used to evaluate future questions. AWS does not identify the unscored items during the exam. Treat all 65 as if they count: spending extra time trying to spot pretest questions is impossible to justify and can damage pacing.

The question formats are multiple choice and multiple response. A multiple-choice item generally asks for one best option. A multiple-response item asks for more than one selection, with the prompt specifying how many choices to select. Read that instruction literally. Selecting only one plausible option when the question asks for two leaves the response incomplete; adding a third speculative option can also make an otherwise sound answer wrong.

The guide does not describe a live coding IDE or a lab in which candidates deploy a real application. The assessment is a question exam. Its scenarios can still require applied development judgment: identify the right SDK behavior, choose a secure credential pattern, diagnose a failed deployment, or select a service feature that meets stated constraints. Practice reasoning from scenario details rather than memorizing console click paths.

What 130 minutes means in practice

Dividing 130 minutes by 65 gives two minutes per presented question. That is a planning average, not a promise that every item takes the same time. A familiar service-selection question might take under a minute. A longer scenario may require comparing constraints, and a multiple-response question can take longer because every option must be evaluated.

A practical pacing rule is to make an initial pass in about 100 to 110 minutes, then leave roughly 20 to 30 minutes for review and questions you flagged. The exact split should be rehearsed in timed practice. If you spend six minutes on a single ambiguous prompt, you have spent the time budget for three other questions. Select the best supported answer, mark it for review if the interface permits, and keep moving.

These checkpoints are coaching targets, not AWS scoring rules. Some candidates prefer a slightly slower first pass and less review; others move faster. A full timed mock shows whether your pace is realistic. If you consistently reach question 55 with only five minutes left, the remedy is usually better recognition of service patterns and a disciplined skip-and-return rule, not faster clicking.

A repeatable question method

1. Read the final sentence first to identify the decision: select a service, explain an error, protect a credential, or choose a deployment behavior.
2. Extract constraints that eliminate answers: least privilege, no long-lived keys, minimal operational effort, low latency, durable state, or a required deployment trigger.
3. Name the relevant AWS mechanism before viewing options when possible. For example, temporary role credentials suggest an IAM role or identity federation rather than embedding a user access key.
4. Check each option for one disqualifying detail. A solution may work technically yet violate security, reliability, or the requested operational constraint.
5. On a multiple-response question, count exactly the requested number of selections and ensure each one is independently supported.
6. If the answer remains uncertain after a focused analysis, choose the best current option and move on. Return only if review time remains.

Worked example: permission error after deployment

An application deploys successfully but receives AccessDenied when writing an object to a designated S3 bucket. The task role has broad read access but no write permission for that bucket. What is the best first correction? A. Put an administrator access key in the application's configuration. B. Add the narrow required write action to the application's execution role for the intended bucket. C. Make the bucket public. D. Retry the deployment until the permission error clears.

The answer is B. The application is already executing under a role, so the failure points to a missing permission in that identity's policy or a conflicting resource policy. Grant only the required action and scope it to the intended resource, then validate any encryption or bucket-policy condition. A creates a long-lived secret and expands risk; C solves neither identity authorization nor confidentiality; D cannot resolve a deterministic authorization denial.

Worked example: repeated event delivery

A serverless handler processes a payment notification. The event source may deliver the same message more than once, and duplicate processing would charge the customer twice. Which design best reduces the effect of retries? A. Assume delivery happens exactly once. B. Make processing idempotent using a stable event identifier and durable record of completion. C. Increase the function timeout. D. Disable all retries without considering failure handling.

The answer is B. Event-driven systems can retry after a timeout or uncertain acknowledgement, so the handler should detect a repeated logical event and avoid repeating the side effect. A confuses at-least-once delivery with exactly-once business behavior. C might reduce timeouts but does not prevent duplicate events. D may discard work during transient faults. A robust design uses idempotency and an intentional dead-letter or failure strategy where appropriate.

Different question shapes to expect

The guide frames the exam around development, security, deployment, and troubleshooting or optimization. Expect scenario questions that combine multiple skills. A code package may be too large because of deployment packaging, an application may fail because a role lacks a permission, a release can require a safe rollback plan, or a latency symptom can come from a poor data-access pattern. The correct answer is usually the smallest change that satisfies all constraints.

AWS may include items tied to emerging development topics, including generative AI, as pretest questions. Because the candidate cannot identify pretest items, answer each question professionally and use the current guide as the boundary of what preparation materials promise. Do not infer that an unfamiliar item signals that the exam is out of scope or that it is safe to skip.

How to use the review pass

Use review to correct an actual reasoning issue, not to replace a defensible answer merely because another option sounds more sophisticated. Re-read the requirement, especially words such as least, only, automatically, persistent, and cross-account. Then check whether the answer meets the stated condition without adding an unsupported architecture. Multiple-response items deserve a final count check.

If time is nearly exhausted, answer every remaining item using the best available reasoning. AWS's exam guide describes a pass/fail result from a scaled score; it does not publish a rule that rewards leaving an item blank. Do not spend the last minutes polishing an already-complete answer while other questions remain unanswered.

Practice under exam conditions

At least a few practice sessions should use a 130-minute clock and a mixed set of original or properly licensed questions. Afterward, review errors by cause: misunderstood requirement, wrong service behavior, IAM misconception, deployment assumption, or careless multi-select. Repeating a question until its answer is memorized is weaker preparation than explaining why the distractors fail and then solving a new scenario.

The exam is delivered through AWS's testing provider, either at an available test center or online with a proctor, subject to location and appointment availability. Follow the provider's current check-in instructions. Test-day identity verification and technical setup consume time before the exam timer begins, so complete those steps early rather than using the scheduled start as a preparation buffer.

Before answering, watch for wording that changes what counts as success. “Most secure” may rule out a solution using a broad user credential; “without changing code” can favor a configuration or managed service choice; “must not process twice” requires an idempotency strategy rather than simply more retries. Under time pressure, underline the actual outcome and one or two decisive constraints mentally, then evaluate each option against them.

A 130-minute mock is useful only if the review is also disciplined. Mark where you are after each quarter of the time, then identify which items caused a delay. If a question took four minutes because you compared every AWS product in the options, study the service boundary that distinguishes them. If it took four minutes because of a missed word, practice careful reading. This diagnosis makes the next timed session more realistic.