In the Netherlands, two longitudinal fieldstudies (N=202 and N=46) were conducted on the relationship between severity of initiation and group attractiveness. Study 1 represents a severe initiation, Study 2 a more mild one. In these two different field `conditions' we aimed to test the dissonance interpretation of the effects of a severe initiation and a positive correlation between severity of treatment of the initiated novices and group attractiveness was expected (hypothesis 1). Further, we expected that feelings of frustration and anger would mediate this relationship (hypothesis 2). On the other hand, our affiliation±attraction hypothesis either predicted a positive correlation between companionship (Rook, 1987) and group attractiveness, irrespective of severity (hypothesis 3), or that companionship would mediate the severity±attraction relationship (hypothesis 4). In both studies, LISREL path analyses only supported hypothesis 3. Contrary to hypothesis 1, severity experienced during the early stages of the initiation correlated with feelings of frustration and loneliness (Study 1), or depressive mood (Study 2). These feelings lasted throughout the initiation and lowered the liking for the group. Results are discussed in terms of newcomers' decisions to leave or to join the group, determined by (i) the exchange and fate control relationship between the leadership and the newcomers; (ii) the severe treatment of newcomers as a selection device and (iii) CCC 0046±2772/97/030275±26