We investigated the feasibility of bioremediation as a treatment option for a chronically diesel-oil-polluted soil in an alpine glacier area at an altitude of 2,875 m above sea level. To examine the efficiencies of natural attenuation and biostimulation, we used field-incubated lysimeters (mesocosms) with unfertilized and fertilized (N-P-K) soil. For three summer seasons (July 1997 to September 1999), we monitored changes in hydrocarbon concentrations in soil and soil leachate and the accompanying changes in soil microbial counts and activity. A significant reduction in the diesel oil level could be achieved. At the end of the third summer season (after 780 days), the initial level of contamination (2,612 ± 70 µg of hydrocarbons g [dry weight] of soil¹) was reduced by (50 ± 4)% and (70 ± 2)% in the unfertilized and fertilized soil, respectively. Nonetheless, the residual levels of contamination (1,296 ± 110 and 774 ± 52 µg of hydrocarbons g [dry weight] of soil¹ in the unfertilized and fertilized soil, respectively) were still high. Most of the hydrocarbon loss occurred during the first summer season ([42 ± 6]% loss) in the fertilized soil and during the second summer season ([41 ± 4]% loss) in the unfertilized soil. In the fertilized soil, all biological parameters (microbial numbers, soil respiration, catalase and lipase activities) were significantly enhanced and correlated significantly with each other, as well as with the residual hydrocarbon concentration, pointing to the importance of biodegradation. The effect of biostimulation of the indigenous soil microorganisms declined with time. The microbial activities in the unfertilized soil fluctuated around background levels during the whole study.