Five methods of absorption rate (Ka) estimation were compared using data from a previously reported bioavailability study: Wagner-Nelson (WN), asymptotic WN (AWN), the Hendeles-Weinberger modification of the WN (HW), nonlinear regression performed on plasma concentration vs time data with fixed elimination rates (NL), and nonlinear regression performed on the cumulative sum of AUCs obtained during WN analysis (AUCNL). The maximum plasma concentration (Cmax) and the time at which Cmax occurred (Tmax) were predicted with each method's respective Ka's, and these were compared to observed values. The WN and HW were consistently biased in their predictions of Cmax and Tmax, providing slower Ka's than any of the other methods. AWN could only be used for the reference treatment. Both NL and AUCNL provided unbiased estimations, but AUCNL had a high level of imprecision. With this set of data, only NL was an acceptable method for accurately estimating Ka.