The purpose of this paper is to provide additional evidence regarding the ability of quarterly time-series models to predict annual net earnings. The degree of accuracy of predictions of annual net earnings is of direct interest to researchers for such purposes as testing models of firm valuation, the relationship between unanticipated earnings and stock price changes, and the information content of voluntary disclosures of annual earnings forecasts by management. In this study, I examine the predictive ability of several different quarterly time-series models. These models include a set of five relatively simplistic ones, firm-specific BoxJenkins (BJ) models, and three parsimonious BJ models. The first section of the paper discusses time-series analysis with particular emphasis on quarterly earnings data. The next section discusses the research on parsimonious BJ models. Remaining sections provide a review of the methodology, including the data, the time-series prediction models, the error measures and hypotheses, the predictive ability results, and finally, a summary and conclusions.