1. Group Ia e.p.s.p.s were recorded from lumbosacral motoneurones in anaesthetized cats after almost complete section of the appropriate dorsal roots. The cable parameters of these same motoneurones were obtained from the voltage response to a brief intracellular current pulse, as described in Iansek & Redman (1973).2. A total of thirty-three e.p.s.p.s, recorded in thirty different motoneurones, were analysed. E.p.s.p.s which were recorded in motoneurones which were not studied using an intracellular current pulse, or in which the resting membrane potential fell below 50 mV, were not considered. Also, e.p.s.p.s whose time course indicated more than one synaptic site of origin were not analysed. The selected e.p.s.p.s were plotted on a semilogarithmic amplitude scale, and their 10-90% rise time, half-width and peak amplitudes were measured.3. Using the previously determined values of the cable parameters L, rho(infinity) and tau(m), the rise time and half-width of each e.p.s.p. were used to determine the synaptic location (X), and the synaptic current time course (alpha). Twenty-seven e.p.s.p.s had time courses which allowed a value of X and alpha to be determined within the constraints of the measured cable parameters. The remaining six e.p.s.p.s either required an extension of the dendritic cable to be localized, or their time course was not compatible with a brief synaptic current.4. The synaptic locations lie in the range 0 (soma) to 1.25 space constants. When expressed as a fraction of the length of the dendritic cable, all but four of the twenty-seven e.p.s.p.s were located on the proximal half of the dendrites.5. The time to peak of synaptic current for each e.p.s.p. ranged from 30 to 390 musec, although a majority (70%) lay in the range 50 to 200 musec. There was no significant correlation between time to peak of synaptic current and synaptic location.6. The peak amplitude of e.p.s.p.s at the soma showed no significant correlation with synaptic location.7. The peak amplitude, and the cable parameters for each e.p.s.p. were used to compute the time course and amplitude of each e.p.s.p. at its point of generation on various fractions of the total dendritic cable, using the results derived in Redman (1973). These calculations showed the greatly increased rate of decay of e.p.s.p.s at their point of generation. Assuming that the synaptic input was restricted to one tenth of the total dendritic tree, the range of peak amplitudes at the synaptic site was from less than 100 muV (soma) to 20 mV.8. The net inward positive charge crossing the synaptic junction was calculated from the voltage-time integral of the e.p.s.p., as was the net outward positive charge crossing the soma membrane. These calculations showed that dendritic synapses caused up to ten times more net charge to be displaced across the synaptic junction than did synapses on or near to the soma, for similar durations of synaptic current. Similarly, dendritic synapses were generally more effective than somatic synapses in displacing charge across the soma membrane. It was concluded that the average quantal content in the conductance change at dendritic synapses is significantly greater than for somatic synapses.9. Some implications of the results for general integrative mechanisms in dendrites are discussed.