Using CDIAC dataset the impact of atmospheric circulation on seasonal fluctuations of carbon dioxide mixing ratios is investigated. For the period 1985-1992 years on average the mixing ratio has increased on 10 ppm approximately and has exceeded 357 ppm for high latitudes of Northern Hemisphere. All sites can be divided into two groups on a degree of CO2 mixing ratio variability during year. The first group includes sites with small annual variability and on some of them (e.g., South Pole or Samoa) the fluctuations almost are absent. The increase of carbon dioxide mixing ratios at sites of this group is associated mainly to the anthropogenic influence. The sites of other group are characterized by large annual variability of CO2 mixing ratios and are located in middle and high latitudes of Northern Hemisphere. The large annual fluctuations may be accounted by three main seasonal factors - anthropogenic, variability of vegetation absorbing CO2, and features of atmospheric circulation providing accumulation or dispersion of CO2. It is possible to consider first of these factors working irrespective of a season. The second factor in high latitudes subjects to the large fluctuations during year as in winter the vegetation is less than in summer, and CO2 is absorbed less also. Confirmation of it can be trend at Alert, Barrow or Mould Bay where the increased values of CO2 mixing ratios are kept during cold months. However, in winter of some years at these sites the two maximums of CO2 mixing ratios are registered and at the neighboring stations they are observed in the same years. It can be associated with features of atmospheric circulation. As an example we have considered the trends since August 1988 till July 1989 at Barrow and Cape Mears. These stations are under influence of teleconnection pattern known as East Pacific (EP) and during strong positive phases of the EP pattern, a deeper than normal trough is located in the vicinity of the Gulf of Alaska / western North America (Barrow), and positive height anomalies are observed farther south (Cape Mears). The considered cold season was characterized by strong positive phases in November 1988 – January 1989 and in April – May 1989 and strong negative phase in February 1989. Such circulating conditions have resulted that in February 1989 maximum of CO2 mixing ratio at Cape Mears corresponded to a minimum at Barrow; the reverse was observed in April. Except for the EP pattern, the North Atlantic Oscillation (NAO), Pacific / North American (PNA) and Polar / Eurasia patterns have an influence on the cold period trends for the stations at middle and high latitudes. The modelling with NCAR CCM3 has shown that under compulsory strong positive and negative phases of the above teleconnection patterns at considered sites it is possible to obtain two maximums in the cold period.