Published critical incident technique (CIT) studies assume that the incidents reported do indeed have an impact on customer–firm relationships, but rarely assess their impact on measures of relationship strength or behavioural response tendencies. In this study, we conducted face-to-face interviews and assessed the impact of critical incidents on measures of relationship strength (i. e. trust and satisfaction). The results confirm that positive and negative incidents have a partially asymmetric impact on measures of relationship strength. Furthermore, we use a multiple indicators, multiple causes (MIMIC) approach to pinpoint which specific incident types have a particularly strong impact on the health of customer–firm relationships. In contrast to published research, our results clearly support the fact that negative incidents, indeed even a single incident, can be severely damaging to a customer– firm relationship and cannot be compensated by corresponding positive incidents.