The familiar rod-drop method for determining delayed-neutron parameters has been refined with new techniques of data collection, analysis, and correction. Values for a highly enriched uranium, heavy-water reactor have been obtained which have a general applicability because they have been accurately corrected for reactor power history, post-shutdown sub-critical neutron multiplication, and finite rod-drop time. Neutron flux after shutdown by rod drop in the Georgia Tech Research Reactor was monitored for periods in excess of three days using two detectors operated in parallel. One detector used a thermal-neutron-sensitive scintillator, the other a fission chamber. Flux-decay data were fit by weighted least squares using the Variable Metric Minimization method. This method was able to fit all the data simultaneously without limit on the number of fitting parameters. The most statistically-significant fit was obtained with 13 delayed-neutron groups, one of which was attributed to background due to its negligibly small decay constant. A fitting expression was used which accurately described the data collection process in which each data point was taken as the time integral of the flux over a finite time interval. The results are compared with values which have been obtained by small irradiated uranium samples and with decay-constant values in the last reported heavy-water in-reactor determination. There are indications that delayed-neutron effectiveness is enhanced by ∼3% in this type of reactor and that the effectiveness of photoneutron groups is decreased by ∼28% because of attenuation of high-energy gamma rays.