The commonly used categorical system for conditionals (zero, first, second, and third) has been repeatedly demonstrated to be inadequate because it is unable to account for many conditional types attested in actual usage. ELT coursebooks and pedagogic grammars are criticized for ignoring the evidence and persisting with its use, but the reality appears to be that, from the point of view of syllabus designers and materials writers, the usefulness and practicality that the system offers outweighs any concerns related to its accuracy. This paper attempts to resolve this tension by showing how the existing system can be improved rather than simply abandoned. It proposes a revised categorization system that better reflects the reality of conditional usage and takes into account evidence on how learners' knowledge of conditional forms expands and develops as they progress through proficiency levels, but that nonetheless can be practically employed in multi-level teaching materials and courses.