The objective of this work was to predict `Packham's Triumph' (Pyrus communis L.) fruit growth as a function of time using an empirical mathematical model. A mature crop was studied at the Experimental Farm of the Comahue National Univ., Rio Negro, Argentina, during the 1992–93, 1993–94, and 1994–95 growing seasons. Trees were selected at random and fruits were collected at weekly intervals. The range of sampling dates was 27 and 178 days after full bloom (DFB). Fresh fruit mass (FM) was measured using an electronic scale (n = 1169). Fruit number/trunk cross-sectional area was also determined; cultural practices were performed according to the local standard program. Equations were developed with SYSTAT procedure. Results showed that the following logistic model provided the most satifactory fit to the pooled data, as compared to the power and linear models: FM (g)= 316.081/(1+ e^5.030–0.039 DFB) R2=0.84 P < 0.001. The accuracy of predictions was tested on an independent crop in the 1995–96 growing season. According to the values of the statistical F test, no significant differences (Pr0.05) were detected between the mean squared deviations of the observed and the estimated values, suggesting that, overall, the model works well. It can provide growers with a means of determining adequate fruit mass at harvest, considering that unless a certain minimum size is obtained, the fruit will be given a lower grade and price.