A method of determining of a true time of arrival of an energy pulse in a positioning system using arrivals times of such pulses at a device after emission of the pulses from respective transmitters. A sampled signal is cross-correlated with the expected form of the energy pulse from the transmitter, and a highest peak in the correlated signal is located and selected to represent a possible time of arrival of the energy pulse. A sequentially preceding peak in the correlated signal is then selected has its magnitude compared with a threshold. This step is repeated until the magnitude of the currently selected peak is less than the threshold, at which point the last-selected peak whose magnitude exceeded the threshold marks the true time of arrival. This process avoids false time of arrival readings caused by reflected pulses.