This paper proposes a liaison -'middle-range realism'-between two long stan
ding explanatory strategies in sociology-'middle-range theory' and 'realist
social theory'. Each offers what the other lacks. Middle-range theory carr
ies an acute sense of the function of theory within empirical inquiry but h
as left undeveloped any notion of its appropriate explanatory form. Realist
social theory has propositional precision but has been unable, in the most
part, to descend from a critical domain to the empirical plane Middle-rang
e realism thus offers a research strategy of the appropriate form and scope
to lead and to federate empirical inquiry. Examples are provided of how mi
ddle-range realism can be applied to improve research using two different s
trategies (survey methods and evaluation research) in two contrasting subst
antive areas (voting behaviour and offender rehabilitation).