We tested the hypothesis that moving the seat of an occupied wheelchair forward would decrease the wheelchair's forward and increase its rear static and dynamic rotational stability, and that moving the seat backward would have the opposite effect. We studied 21 able-bodied subjects in a lightweight, manually propelled wheelchair with the occupied seat in the normal position, 5cm forward, and 5cm backward. Regression analysis revealed that, within the range studied, the relationship between seat position (chi, in cm) and the forward and rear static stability (SSf and SSr, in degrees) could be expressed by the linear equations SSf = 15.52 (0.06)-0.70 (0.01) chi (r2 = 0.98) and SSr = 24.55 (0.07) + 1.11 (0.02) chi (r2 = 0.99). These findings were consistent with the results of a computer model that we used to predict the nature and extent of the effect on static stability, except that the model showed that the relationship becomes nonlinear with seat-position changes greater than those we studied. The relationship between seat position and dynamic forward and rear stability (DSf and DSr, the threshold speed [in m/s] required to induce a transient tip) could be expressed by the equations DSf = 0.74 (0.01)-0.041 (0.002) chi (r2 = 0.91) and DSr = 0.63 (0.01) + 0.030 (0.002) chi (r2 = 0.91). The results support our hypothesis and indicate that seat position is a very significant determinant of wheelchair stability.