Waterfall Model
Sequential development phases with each completed before the next begins.
The Waterfall Model is the sequential software-development model in which work proceeds through phases (typically Requirements, Design, Implementation, Verification/Testing, Maintenance) with each phase completed and signed off before the next begins, and downstream changes flowing back upstream only with substantial cost. The model's most cited origin is Winston Royce's 1970 paper 'Managing the Development of Large Software Systems', though Royce's actual paper presented sequential development as a problematic baseline that should be modified with feedback loops, prototyping, and iteration — the deficiencies of pure waterfall were Royce's argument, not his prescription. The waterfall framing nonetheless became the standard model for large software projects through the 1970s, 80s, and 90s, particularly in defense and government contracting where DOD-STD-2167 and similar standards effectively mandated waterfall-style processes. The Standish Group's CHAOS reports beginning in 1994 documented the high failure rates of waterfall-style projects, contributing to the rise of agile alternatives. Waterfall remains in use in some regulated industries (medical devices, aerospace, government) where formal phase-gates align with regulatory requirements, but is widely regarded as a worst-fit-for-most-software-projects model in modern practice.
Core components
- Sequential phases: Requirements, Design, Implementation, Verification/Testing, Maintenance (variant phase counts)
- Each phase completed before next begins
- Phase-gate sign-offs
- Documentation as primary artifact
- Limited iteration or feedback between phases
- Royce's actual paper proposed feedback loops that pure waterfall ignores
- DOD-STD-2167 codification
- Decline through 1990s-2000s as agile alternatives emerged
Primary use case
Large software projects in regulated industries (medical devices, aerospace, defense, government) where formal phase-gates align with regulatory requirements; legacy organizational contexts; teaching framework as the canonical alternative to agile; foundational reference for understanding what agile is reacting against; some hybrid models combine waterfall structure with internal agile sub-projects.
Common criticisms
- Royce's original 1970 paper actually critiqued pure sequential development and proposed feedback loops — the 'waterfall model' as taught and practiced was a misreading of Royce
- the model assumes requirements can be fully understood upfront, which rarely holds in practice
- late discovery of requirements problems is enormously expensive when previous phases are 'complete'
- documentation produced in early phases goes stale before later phases use it
- phase-gate sign-offs become bureaucratic rituals that don't actually verify quality
- Standish CHAOS reports documented persistent high failure rates
- the model's principal mainstream use is now as the foil that explains why agile exists
- even regulated industries have substantially adapted waterfall with iterative practices in modern interpretations.
Lineage
- Siblings
- Spiral Model