Previous studies examining environmental context-dependent memory (ECDM) effects using indirect measures of memory have produced inconsistent findings. We report three experiments that examined ECDM in an indirect memory paradigm (word-fragment completion) using ambient odours as environmental contexts. Expt 1 manipulated the odour present at learning and testing (rosemary or lemon) to produce reinstated-context or switched-context conditions. Reinstating rosemary led to a striking ECDM effect, indicating that indirect memory testing can be sensitive to ECDM manipulations. Odour ratings also indicated that rosemary induced a more unpleasant mood in participants than lemon. Expt 2 assessed the influence on indirect retrieval of odour-based mood induction as well as odour distinctiveness, and indicated that rosemary's capacity to promote ECDM effects appears to arise from an additive combination of its unpleasantness-inducing properties and its distinctiveness. Expt 3 partially supported these proposals. Overall, our findings indicate that some odours are capable of producing ECDM effects using indirect testing procedures. Moreover, it appears that it is the inherent proprieties of odours on dimensions such as unpleasantness and distinctiveness that mediate the emergence of ECDM effects, thereby explaining the particular potency of rosemary's mnemonic influence when it is reinstated.