Résumé : In order to investigate the reliability of the endogenous concept of depressive illness with some sleep EEG parameters, we studied 39 male inpatients suffering from a nonbipolar major depressive episode (15 endogenous (MDDE) and 24 nonendogenous (MDDNE)) and 20 age and sex matched normal controls (C). All patients were diagnosed according to the Research Diagnostic Criteria (RDC) and the endogenous character of the episode was assessed with the Newcastle Endogenous Depression Diagnostic Index. We found significant differences for the following variables between the three groups (MDDE, MDDNE and C): sleep period time (SPT), REM latency, stage II, slow wave sleep (SWS), REM latency expressed as a continuous variable and REM latency expressed as a dichotomizing variable with a threshold of 50 min. These variables were used to compare the endogenous and the nonendogenous depressed patients and also the major depressed patients and the normal controls. Significant differences were observed between all depressed patients and control subjects for amount of SWS and REM latency which were both reduced in endogenous and nonendogenous depressed patients. No significant difference was observed between endogenous and nonendogenous depressed patients, except for the REM latency expressed with a threshold of 50 min (more frequently observed in endogenous depressed patients). Our data support the observation that SWS and REM latency are decreased in major depressive patients. However, in this age and sex controlled study, subtyping nonbipolar major depressive disorder for an endogenous character by the Newcastle Endogenous Depression Diagnostic Index (NEDDI) did not reveal further significant differences for sleep EEG variables, except for the shortening of the REM latency expressed as a dichotomizing variable.