Field experiments were conducted with nine treatments for soil application, having three magnesium sources: ESTA® Kieserite (kieserita), magnesium sulfate of agricultural use and magnesium oxide, with three doses each, 30, 60 and 90 kg ha^-1 in terms of magnesium oxide, two treatments having a soil application of kieserite, 60 kg ha^-1 plus foliar application of Epsom salt and a last one having 60 kg ha^-1 of magnesium oxide plus sulfur. There was one check treatment with no magnesium added. As experimental design completely randomized blocks with three replications were arranged. Two potato varieties, the “Parda Pastusa” variety which is consumed as fresh potato and the “Diacol Capiro” variety which is consumed as processed potato in the industry, were used. The experiments were developed in farms of four localities of the Andean Colombian highlands, where potato is the main crop and whose soils were classified as Typic Melanudands (Villapinzón locality), Typic Dystrudepts (Siachoque locality), Andic Dystrudepts (Toca locality) and Typic Haplustepts (Samacá locality). The main objective of this study was to gather more detailled information of the nutrient requirement of potato varieties and availability of different magnesium sources in order to derive practical recommendations on fertilization strategies for farmers in the Colombian highlands. In Villapinzón the highest yield of potato tuber for the Parda Pastusa variety was of 46.4 t ha^-1 as a result of the treatment composed by 60 kg ha^-1 of magnesium oxide plus sulfur, compared to the check treatment of 41.7 t ha^-1. In Siachoque, the highest yield was of 47.9 t ha^-1 under the same treatment, with a check treatment of 34.8 t ha^-1. The previous treatments also leaded to the highest magnesium absorption in the whole plant (root, steam leaves and flowers) 0.92 grams per plant compare to the check treatment of 0.69 grams per plant n both localties. On the other hand, the Capiro variety produced a yield of 47.1 t ha^-1 under the treatment of 30 kg ha^-1 of kieserite compare to the check treatment of 43.4 t ha^-1 in the Toca locality, while 36.4 t ha^-1 of potato tuber were obtained in the Samacá locality, under the treatment of 60 t ha^-1 of kieserite, the yield of the check treatment was of 31.5 t ha^-1. In relation to the tuber quality, the kieserite treatments both 30 and 60 kg ha^-1, produced the highest dry matter content in tuber, 25.8% (check treatment 22.1%) and 26.6% (check treatment 23.1%) in the localities of Toca and Samacá respectively. Reducing sugars showed their lowest value of 0.16% under the treatment of 90 kg ha^-1 of magnesium oxide (check treatment 0.20%) in Samacá. A value of 0.07% of reducing sugars for the treatment of 60 kg ha^-1 plus sulfur was found in Toca (check treatment 0.17%). The previous treatments showed correlation with other variables like sprout percentage, foliar area index, chlorophyll content in leaves, calcium, magnesium, potassium and sulfur content per plant and tuber size. It can be concluded that magnesium oxide in doses of 60 kg ha^-1 was the best treatment for the Parda Pastusa potato variety (fresh consumption) and kieserite in doses of 30 and 60 kg ha^-1 were the best treatments for the Diacol Capiro potato variety (industrial use) for most of the variables studied.