The results of controlled environment experiments and a field survey covering six major New Zealand kiwifruit (Actinidia deliciosa) growing regions over 3 years showed, surprisingly, that the effect of temperature on the rate of fruit growth is small, at least during the second half of the fruit growth period. The considerable variation in the mean and standard deviation of fruit volume at harvest observed in the field among seasons and sites is therefore not attributable to temperature differences during the fruit-growing season. This raises the possibility that most of the factors affecting fruit growth rates may be established early in the season, so harvest fruit volumes can be predicted from early-season measurements. Mean fruit volumes observed in the field survey ranged from <85 ml (Kerikeri 1987 and Te Puke 1989) to > 130 ml (Kerikeri 1988 proportion of this should be attributed to management practices such as fruit thinning. The fruit growth curves could be described by a two-phase curve, with a reduction in slope 50-60 days after flowering. The initial phase, up to 50 days from flowering, produced growth rates of 1.55 ml/day. The second, slower phase of growth, averaged 0.42 ml/day over the period to harvest. A simple linear regression model was used to predict the mean fruit volume at harvest from measurements of fruit volume made at a given time during the period from 50 to 110 days from flowering (typically mid January and mid March, respectively). The percentage of the variation in mean fruit volume at harvest accounted for by the regression was 75% when based on measurements made 50 days from flowering and rose to nearly 98%, 110 days from flowering. A similar approach was used to predict the standard deviation of fruit volume at harvest. The observed standard deviation at harvest varied from 10 ml to just over 20 ml, which is within the range published by others. The regression of the standard deviation of the fruit volume at harvest on the standard deviation of fruit volume earlier in the season accounted for more than 80% of the observed variation from 50 days after flowering onwards.