Part of a larger research program studying communication in the cockpit, this study investigated the occurrence of mitigation, that is, the use of linguistic forms that convey propositional content without giving offense, and its relationship to aviation accidents. The data were from flight-recorder accident transcripts containing observable degradation or failure of crew coordination that was actually or potentially critical to flight completion. Initial analysis of the data suggested several hypotheses concerning the use of mitigation, and quantitative measures were developed and validated. The findings supported the five hypotheses that: (1) requests to superiors are more mitigated; (2) requests are less mitigated in crew-recognized emergencies; (3) requests are less mitigated in crew-recognized problems; (4) topic-failed speech acts (i.e., those not followed up immediately) are more mitigated; and (5) crew suggestions unratified by the captain are more mitigated. A proposed classification of possible contexts for communication, divided according to measures for communicative effectiveness, includes four general contexts (operational, pedagogical, rhetorical, and social/relational) and a number of subcontexts for application in discourse analysis and communicative effectiveness. (MSE)