Potato response to environment, planting date and genotype was studied for different agro-ecological zones in Lesotho. Field experiments were conducted at four different sites with altitudes ranging from 1,655 to 2,250 m above sea level during the 2010/2011 and 2011/2012 summer growing seasons. Treatments consisted of three cultivars that varied in maturity type, two planting dates and four sites differing in altitude and weather patterns. Various plant parts were measured periodically. To understand and quantify the influence of abiotic factors that determine and limit yields, the LINTUL crop growth model was employed which simulated potential yields for the different agro-ecological zones using weather data collected per site during the study period. Observed actual crop yields were compared with model simulations to determine the yield gap. Model simulations helped to improve our understanding of yield limitations to further expand potato production in subtropical highlands, with emphasis on increasing production through increased yields rather than increased area. Substantial variation in yield between planting date, cultivar and site were observed. Average tuber dry matter (DM) yields for the highest yielding season were above 7.5 t DM ha −1 or over 37.5 t ha −1 fresh tuber yield. The lowest yield obtained was 2.39 t DM ha −1 or 12 t ha −1 fresh tuber yield for cultivar Vanderplank in the 2011/2012 growing season at the site with the lowest altitude. Modelled potential tuber yields were 9-14 t DM ha −1 or 45-70 t ha −1 fresh yield. Drought stress frequently resulted in lower radiation use efficiencies and to a lesser degree harvest indices, which reduced tuber yield. The site with the lowest altitude and highest temperatures had the lowest yields, while the site with the highest altitude had the highest yields. Later maturing cultivars yielded more than earlier maturing ones at all sites. It is concluded that the risk of low yields in rain-fed subtropical