The high efficiencies of large volume 'coaxial' Ge(Li) detectors make them attractive for coincidence experiments. However the electric field in this configuration is not uniform and hence the charge collection times vary widely as a function of position in the detector. The resultant variations in output pulse shape limit coincidence resolving times. In our studies one 511 keV annihilation quantum (22Na source) was detected in a scintillation counter to establish the zero (start) time and its partner in the Ge(Li) detector (at 1500 V) was accepted only when a full energy pulse was registered. The stop pulse to the timesorter was generated by the leading edge of the Ge(Li) preamplifier pulse at ~ 10% (50 keV) of maximum amplitude. Scanning the detector with a 1 mm wide collimated Â¿-beam resulted in time distributions whose full widths at half-maximum (fwhm) were ~ 8 nsec in the coaxial region of the detector broadening to ~ 35 nsec at the closed end. The distributions are skew in shape with long approximately exponential tails having half value slopes of ~ 5 to ~ 20 nsec. Similar studies with stop pulses generated at ~ 50% of maximum amplitude show much broader time distributions. The position variation in pulse shapes due to charge collection time was confirmed from oscilloscope photographs by triggering with the scintillation counter 'start' pulse.