A combination of objective and subjective variables was used to test a version of the withdrawal progression model. The sample consisted of 200 public sector employees for whom objective lateness and absence data were available for two time periods. Data were analyzed via the CALIS procedure for analyzing structural models. The best fitting model for predicting satisfaction and intention to leave was attained when lateness and absence in two consecutive years were used as predictors. Competing models that included demographic and environmental factors showed poorer fit. The findings support Bem's self-perception model (1972) or Clegg's proposition (1983) that lateness and absenteeism reduce satisfaction through their effect upon tangible or social sanctions.