AbstractWe compiled data from the Swiss seed certification programme for the country‐wide incidence of viruses in seed potato crops for the years 1989–2012. Model selection techniques were used to regress year‐to‐year variation in the incidence of potato viruses – largely dominated by Potato virus Y (PVY) – in three susceptible varieties against the abundance of virus vectors (winged aphids), obtained in a suction trap, to identify the most important vector species. The ultimate aim of this study was to develop a decision‐support system capable of forecasting virus spread during the current season using trap data of aphid flights. The average virus incidence in the varieties Bintje, Sirtema and Charlotte varied considerably among years, ranging from 1.0% in 2009 to 13.6% in 1989 (N = 150–611 seed lots per year). A linear regression model including the cumulative sums (until mid‐June) of two aphid species (Brachycaudus helichrysi and Phorodon humuli) as predictor variables for virus disease was remarkably well supported by the data (R2 = 0.86). Similarly, using counts of B. helichrysi alone resulted in a good model fit (R2 = 0.81). Cross‐validation revealed high predictive accuracy of the model. Although prediction root mean squared errors (RMSE) calculated for different timings of forecasts were high for extremely early forecasts, they rapidly declined for forecasts conducted by the end of May (i.e. 2–4 weeks after potato emergence). Winter temperature (January–February) was positively correlated with the abundance of B. helichrysi in early summer as well as with post‐harvest virus incidence. Remarkably, the abundance of Myzus persicae, often considered the main vector of PVY, was not correlated with virus incidence. Taken together, our analysis suggests that the early migrating aphid B. helichrysi, rather than M. persicae, is the main vector of PVY in Switzerland, and that suction trap data are useful for the design of decision‐support systems aimed to optimise virus control in seed potato production.