Purpose - This research seeks to explore the factors predicting customer loyalty in retail banking. Loyalty was measured in terms of a customer's willingness to recommend a bank and their intention to remain with their main bank short-term (in the next six months) and long-term (from six months to five years). Design/methodology/approach - The study was based on a mail survey of 1,951 individuals. Potential predictors were drawn from the literature and included in three separate regression models to model different types of loyalty. Findings - The results indicate that willingness to recommend is best predicted by affective attitude, overall satisfaction and empathy. Short-term behavioural intentions, however, were best predicted by overall satisfaction and responsiveness, while long-term intentions were predicted by overall satisfaction, affective attitude and empathy. The three models explained a substantial amount of the variation in the dependent variables: 71 per cent for willingness to recommend, 43 per cent for short-term intentions and 46 per cent for long-term intentions. Research limitations/implications - The study adds to the discussion of the relationship between perceived satisfaction, service quality and a customer's intentions to recommend a bank and/or remain a customer. The results also contribute to the development of more parsimonious models, suggesting that affective attitude, overall satisfaction, empathy and responsiveness together explain a large percentage of the variation in customers' intentions. Practical implications - Based on this study's findings, banks can profile customers with potential for defection based on only four variables. Originality/value - The results demonstrate the importance of satisfaction measures and some SERVQUAL dimensions in predicting loyalty in retail banking. It also found evidence that not all five SERVQUAL measures are needed to profile customers and predict loyalty.