In the studies on climate and its changes in the polar regions it is essential to determine climatic seasons which can be based on thermal, circular and phenological criteria but also according to different types of weather. The aim of this research is to determine thermal seasons, to characterize their structure and general regularities, which may make the more detailed environmental monitoring of these areas possible. According to many authors, a climatic characterization of a given area should be presented through defining its seasonal structure. This article attempts to find natural thermal periods in the polar climate, which differ from the conventional, fixed monthly or quarterly periods: spring III-V, summer VI-VIII, autumn IX-XI, winter XII-II; often accepted by many scientists in order to make the characterization of the course of selected meteorological elements easier. The analysis of the seasonal structure of the climate of Hornsund is based on the data from the period of 1990-1999. The indices that characterize the initial and final dates, the overall duration of the thermal seasons, and estimation of the seasons? changeability in thermal terms have been taken into consideration. Calendar boundaries have been set according to the method proposed by Kosiba (1958), in which the date that begins the period of the domination of days with the daily average air temperature (Ti) typical for a given season is accepted as the season?s boundary. As the quite significant changeability of the daily average air temperature complicates the choice of initial and final dates of seasons, additional criteria are used: the number of days proper for thermal season (w), days warmer than w?days, days colder than w-days, the average air temperature and other. This study provides a division into four seasons according to Baranowski?s criteria (1986) accepted on the basis of an analysis of the annual course of air temperature in Hornsund, the accepted thermal criteria are as follows: spring -2.5°C <= Ti <= 2.5°C, summer Ti >= 2.5°C, autumn -2,5°C <= Ti <= 2.5°C, winter Ti <= -2,5°C. The characteristics of a vegetative period are also defined. Its duration in the polar regions is difficult to estimate. If we accept the most commonly used criterion of the stabilization of the daily average air temperature over +5°C, we will face the situation in which the vegetative period in the polar regions is either very short or does not occur at all. Phenological observations of Sorkappland - S Spitsbergen (Dubiel, 1988) made it possible to estimate a natural thermal threshold 0?C which begins the vegetative period. The development of most plants and their first flower buds occurs in average air temperature of approximately 0?C. Blooming and producing seeds, on the other hand, occur when the air temperature exceeds 2.5°C. Seasons (fig. 17), determined on the basis of daily average air temperatures, characterize and emphasize the changeability of thermal conditions and the specifics of the polar climate very well, what results in the conclusions enumerated below: - in the researched decade the initial and final dates and the overall duration of the thermal seasons are characterized by great changeability, - the most stable, with regard to the initial date, are spring and summer, - the most changeable, with regard to duration, are autumn and summer, - the most thermally stable season is summer. The least thermally stable season is autumn, - transitional seasons have a tendency to prolong: mainly autumn (the effect is that winter becomes shorter) and to a lesser extent spring. Winter and summer shorten, - the analysis of the line of this trend reveals that summer gets slightly colder. Spring and winter do not show any significant changes, - the most visible tendency is a downward tendency of autumn temperatures - the effect of the prolonged duration towards winter, - a vegetative period shows a tendency to begin later and to finish slightly earlier. The final date, however, does not reveal any significant tendency for changes.