Purpose: To examine the effects of specific personal and job characteristics on year-to-year (2000-2001) changes in teachers' frequency of absences. Design/methodology/approach: With few exceptions, the population of elementary- and middle-school teachers in the Israeli public education system (N=51,916) was studied. Hierarchical regression analysis was used. Findings: Prior absenteeism, age, education, and supervisory position were found to be significant predictors of absenteeism frequency, accounting for about 50 percent of the variance in absence frequency. Originality/value: This study focuses on relatively stable individual-difference predictors, including sociodemographic variables and work-related characteristics, which have been downplayed in the literature. These predictors can be measured more reliably and validly, compared to complex psychological constructs, and are relatively easy to interpret and implement by decision makers. (Contains 2 tables.)