A method to estimate coffee production based on flowering dates was evaluated. The flowering events were registered in two periods corresponding to the first and second semester harvests in four commercial plots of Colombia variety coffee. To register each event two simultaneous sampling methods were used: furrow and random. The flowering registry in each selected coffee tree consisted of taking the branch with the greatest number of flower buds at the preanthesis stage in order to count them and the number of branches with buds in that stage. The fruit set percentage and the weight per cherry were assessed and the coffee cherry production in kilograms was registered. Using that information, the percentage of trees with flowers, total of flowers, flowering percentage and production were estimated. In the estimation of the proportion of flowering trees there were absolute errors under 5%; the relative error rate for the estimation of the number of flowers was 13% for the flowerings corresponding to the first flowering period and 13.7% for the main harvest. The errors for production were ≤20% for 44.1% of the events when the furrow sampling method was used and 35.3% of the events in the randomized sampling. The minimum sample size to estimate the mean number of flowers per tree/plot is 250 trees, which corresponds to a relative error of less than 20%.