Purpose For firms listed on the New Zealand Stock Exchange, which is a relatively thinly traded market, the purpose of this paper is to examine the nature of stock returns associated with a dividend omission announcement when computations specifically address thin trading, and whether specific firm characteristics affect the likelihood and nature of a dividend omission. Design/methodology/approach First, event study analysis is used to check if dividend omissions actually do impact share prices in terms of short-term abnormal returns and longer-term cumulative abnormal returns (CARs) in a thinly traded market. Second, binomial logistic regression analysis is used to determine what, if any, company characteristics are associated with the decision to omit a dividend. Third, multinomial logistic regression analysis is employed to determine what firm characteristics are associated with continuing (or ending) a phase of no dividends before a dividend resumption. Findings Dividend omissions generate immediate negative abnormal returns, and there is a longer-term persistence of negative CARs. The size and duration of these abnormal returns are smaller, but still significant, when thin-market-specific methodology is employed. With respect to firm characteristic, smaller firms, firms with decreased earnings, a higher level of extraordinary charges, greater leverage and firms with a higher book-to-market value are associated with a greater likelihood of making an omission. With respect to the length of time between an omission and resumption of dividend payments, earnings decreases, a higher book-to-market value, a higher level of extraordinary charges and a decrease in firm debt level become significant. Originality/value This paper adds value in two dimensions. First, it considers dividend omissions in three different, but inter-connected ways. Second, the use of multinomial logistic regression to examine an aspect of the non-payment hiatus breaks new ground.