Abstract: When someone makes research on near-Earth satellite orbit determination, it is inevitable that one should use one of the atmospheric models, such as exponential atmospheric model(EXPT hereAfter), DTM, CIRA86, J77, and so forth. As a compara tively later established atmospheric model, J77 seems so conspicous that it is usually taken into consideration as far as air drag perturbation is concerned. Here, the author is involved in a puzzle of calculation divergence resulting from introduction of J77 as an atmospheric model to deal with the precise orbit determina ton of the near-Earth satellite Ajisai when a computer operating system at themicro-VAX3800 is utilized. Based on many simulating calculations and postanalyses, the author has inferred that calculation divergence may be brought about by improperly adopting a default value of parameterXin our software SHO RDE for research on satellite orbit determination. The calculation shows that this parameter is dependent on, but not just equal to, the significant value of a specific computer operating system. After adjusting the parameter X, the author recalculates the orbital elements and find out the results convergent. Finally the author compares the results with that produced by other atmospheric models, and the comparison indicates that the adjustment of parameter Xis rational.