Abstract Kiwifruit ( Actinidia deliciosa (A. Chev.) C. F. Liang & A. R. Ferguson) is a perennial vine which requires winter chilling during the dormant period before it will flower and crop adequately. Quantifying the impact of winter temperatures is difficult because the first detectable responses of the vine are floral initiation and bud break which do not occur until spring. We believed that the respiration rate (R D ) of buds of kiwifruit during the dormant period might provide an early indication of the changes in bud state associated with chilling. R D was measured during autumn, winter and spring on buds collected from three contrasting regions of New Zealand. The R D of buds collected from these regions during the autumn were consistently different. Buds from the coolest region had the highest R D . These differences which were small, but consistent, were apparent by the time measurements started at the beginning of May. The autumn R D was correlated with the proportion of floral shoots and numbers of flowers per winter bud, and to a lesser extent, the amount and spread of bud break. These correlations focus attention on the importance of this period in determining ‘winter’ chilling responses. It raises the possibility of R D being used as a predictor of vine performance as early as 17 weeks before bud break. A rapid increase in R D was detected during spring. This occurred 3–6 weeks prior to bud break and coincided with developmental changes within the buds. Changes in R D in early spring were correlated with the timing and spread of bud break and the proportion of shoots that bore flowers, but not flower numbers. To provide a comparison of the potential use of R D as an index of vine response to chilling, bud break and flowering parameters were regressed against mean air temperature during the period from May to July. The correlations with air temperature were as useful as those based on the autumn or spring R D but were available only 6–9 weeks before bud break. The autumn R D can be obtained at least 8 weeks earlier than mean winter temperature data and this may be an important advantage if this information can be used for forecasting vine responses to chilling.