Threat modeling is the systematic process of identifying, classifying, and prioritizing a system's potential threats before they materialize into actual vulnerabilities. Unlike penetration testing, which happens after a system is built, threat modeling ideally occurs at the design stage — precisely where fixing an architectural flaw is cheapest. Three widely used methodologies in this space take fundamentally different approaches.
STRIDE: classification-driven
STRIDE, developed at Microsoft, divides threats into six categories: Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege. Its main strength is simplicity and fast onboarding: a development team draws a data flow diagram and, for each element (process, data flow, data store), asks the six STRIDE questions. Its weakness is that STRIDE alone provides no risk prioritization — it simply generates a list of possible threats that must be ranked separately (typically with DREAD or CVSS).
PASTA: business-risk-driven
PASTA (Process for Attack Simulation and Threat Analysis) is a seven-stage framework that explicitly ties technical threats to business objectives. Unlike STRIDE, which starts from the system's architecture, PASTA begins by defining business objectives and their associated risks, then moves toward identifying technical assets, vulnerability analysis, and attack simulation. This approach carries more weight for organizations that need to justify security decisions to senior leadership — who speak the language of business risk, not technical minutiae — but its time cost and implementation complexity are considerably higher than STRIDE's.
Attack trees: scenario-driven
Attack trees, formalized by Bruce Schneier, model threats as a tree structure: the attacker's ultimate goal sits at the root, and each branch represents a possible method of achieving that goal, which can be decomposed into finer sub-branches. Their key advantage is the ability to model multi-stage attack chains — exactly what happens in real-world attacks, where an adversary combines several low-severity vulnerabilities to reach a high-severity outcome. Their weakness is poor scalability for very complex systems; attack trees are more practical for a specific feature or subsystem than for an entire enterprise platform.
Which method should you choose?
In practice, these three methods are not competitors but complements. A mature approach might use PASTA to prioritize risk at a high level based on business impact, then apply STRIDE to each sensitive technical component for systematic coverage, and finally draw attack trees for the highest-risk, most complex attack scenarios. Choosing a methodology is itself an engineering decision that should be driven by organizational maturity, the audience for the final report, and the complexity of the target system.