Nuclear magnetic resonance measurements of a porous medium using multiple wait times are used for determining the longitudinal relaxation time T.sub.1 time of oil in a hydrocarbon reservoir. The T.sub.1 values of water in the formation are represented by a distribution. In a dual wait time implementation of the method, data are acquired with a short wait time TWS chosen such that the wetting fluid phase (e.g., brine) is completely relaxed but the hydrocarbon phase is partially relaxed, giving a sequence ECHOA and with a long wait time TWL giving an echo sequence ECHOB. Using known methods, the T.sub.2 for the oil and water may be determined. The individual echo trains are summed and the T.sub.1 values determined from the summed echo trains and the determined values of T.sub.2. The summation greatly improves the SNR. An equivalent method determines the T.sub.1 values by summation of a first echo train that is the sum of the individual echo trains and the summation of a second echo train that is the difference of the individual echo trains. The various summed values may also be used as for quality control of the data.