To gain protection under the International Convention for the Protection of New Varieties of Plant, new plant varieties must be distinguishable from existing varieties in at least one important characteristic. Assessment of quantitative characteristics often uses a procedure based on analysis of variance of variety-by-year means for two years of trials. In the current paper, a new method is described that can identify those reference varieties that are so different from a candidate that there would be no reason to compare them in the subsequent year, resulting in potential cost savings. It is more objective and transparent than existing practice for quantitative characteristics based on expert opinion. The method calculates thresholds for quantitative characteristics. The thresholds are defined so that if in the first year the difference between two varieties in a characteristic is larger than the characteristic's threshold then it is highly likely that the varieties would be distinct after two years. Thresholds were derived based on statistical predictions of the full decision after two years using the first year results combined with historical data. It is shown that these thresholds are sensitive to yearly heterogeneity in the variety-by-year variation. The method accommodates this heterogeneity by modelling yearly residual variances with the inverse gamma distribution. This extension meant that exact analytic formulae were not available so an approximation was suggested. Using simulation it was found that the approximation was reasonable; for thresholds corresponding to a high probability of distinctness, the approximate thresholds were a little higher than required. The method was evaluated on a 19 year data set for field pea, comparing decisions based on first year thresholds with those based on the full two years. It was found that with the probability of distinctness set at 0.99, the calculated thresholds were generally lower than the existing expert-set thresholds but had acceptable levels of false positives and false negatives.