Rational action theory (RAT) is not a highly unified intellectual enti
ty. In the first part of the paper, varieties of RAT are distinguished
in terms of three criteria: i.e. according to whether they (i) have s
trong rather than weak rationality requirements; (ii) focus on situati
onal rather than procedural rationality; (iii) claim to provide a gene
ral rather than a special theory of action. In the second part, these
same criteria are applied in a consideration of which version of RAT h
olds out most promise for use in sociology.