We analyze here the behavior of the magnitudes of the F 1 and E peaks of the electron density profiles measured by the Radio Science Subsystem of the Mars Global Surveyor spacecraft, as a function of solar zenith angle ฯ and solar flux. For each of the 658 days of data in the six occultation seasons in the northern hemisphere, we choose one profile to analyze, which is that for which the F 1 peak is the median value. We assume that the variations of the measured peak densities can be represented as A(cos ฯ ) a and as B(F 10.7 ) b , where F 10.7 is the usual solar flux proxy, appropriately shifted to the orbital position of Mars. To minimize the effect of solar activity, we divide the data into 6 F 10.7 bins, fit the data in each bin, and derive the values of the exponent a and the coefficient A F 10.7 for each bin. The median values that we derive for the exponent a is 0.46 for the F 1 peak, and 0.395 for the E peak. To minimize the effect of SZA, we divide the data into eight SZA bins, and derive the exponent b and the coefficient B ฯ for each SZA bin. We argue that the last three SZA bins should be excluded because the fits were poor, due partly to the small number of data points in each of these bins. If we do so, the median values of b that we derive are 0.27 and 0.40 for the F 1 and E peaks, respectively. Finally we derive a 3-parameter fit to all the data, which expresses the variability of the peak densities as a function of (cos ฯ ) a and (F 10.7 ) b simultaneously. The fitted values of the exponents a and b for the F 1 peak are 0.45 and 0.26, respectively; for the E peak, the values are 0.39 and 0.46, respectively. We compare our results to Chapman theory, and to those of other investigators.