Continuous self-monitoring (CSM) was compared with a demand characteristics control condition (non self-monitoring), with intermittent self-monitoring (ISM) and with another control condition. It was predicted that both self-monitoring conditions would produce effects over and above the demand characteristics inherent in the self-monitoring procedure. It was also hypothesized that ISM contains a partial (self) reinforcement component; thus it was postulated that ISM would be superior to CSM with regard to the durability (resistance to extinction) of reactive effects. The target behavior for suppression was speech interruptions (SI). All groups were requested to give impromptu speeches on familiar topics during three periods: (1) 5-minute baseline, (2) 12-minute self-monitoring phase and (3) 12-minute follow-up. The results showed that continuous self-monitoring has significant effects over and above the suppressive effects of demand characteristics. Reliability checks confirmed previous findings that the accuracy of self-monitoring is low. CSM proved to be significantly superior to ISM and both control conditions for decreasing SI and maintaining the decrease. ISM was considered to be a more difficult task which disrupted speech fluency. CSM subjects also received greater practice in noticing and suppressing SI due to the longer time they spent monitoring SI. It was speculated that the partial reinforcement effect may not be applicable to self-reinforcement. (Author)