Recent studies show advancing onset of plant growing season in many regions for the last several decades. With the well-established dependence of plant phenology on temperature, these trends are interpreted as an indication of global warming. For several decades, however, other determinants of plant phenology, e.g. varieties and trends in managed systems, may have changed and confounded the phenological trends. In this study, we tested if long-term changes in phenology of apple (Malus pumila var. domestica) are attributable to long-term changes in temperature by comparing the phenological response to long-term trend in air temperature, which is of our interest, with that to year-to-year fluctuation in air temperature, which should represent the real effect of temperature on phenology. We collected records of air temperature and phenological events (budding and flowering) in apple from 1977 to 2004 at six locations in Japan. Linear trends in flowering showed advancing rate in the range from 0.21 to 0.35 day yr−1, statistically significant at three locations (P<0.05). We also found a warming trend in mean air temperature throughout March and April, with which flowering was closely correlated, in the range from 0.047 to 0.077 °C yr−1, statistically significant at five locations (P<0.05). We separated the temperature time-series into two components: a long-term trend and a year-to-year fluctuation, by fitting smoothing spline to the trend and taking the residuals as the anomaly. We then fit a multiple regression model of phenological response to air temperature with separate coefficients for long-term trend and anomaly. Flowering date responded to the long-term trend at −3.8 day °C−1 and to the anomaly at −4.6 day °C−1. The temperature coefficients were not statistically different from each other or among locations, suggesting that the advance of apple phenology has predominantly been caused by the temperature increase across the locations studied. The same result was also observed with budding.