An estimation of the critical period of weed control is helpful in formulating appropriate weed-control strategies. A regression approach is presented to estimate the thresholds of critical period of weed control and time of equal interference (or time of onset of competition). In this approach, yields were either a linear or logistic function of the duration of weed-free and weed-infested periods. Confidence intervals of the thresholds of critical period and time of equal interference were determined for the linear model. An approximation to the standard error of critical period and associated confidence interval were given for any general form of the model. The method was applied to estimate the critical period of weed control in rainfed lentil using data from four field experiments conducted in Jordan. The relationship of yield with the duration of weed-free period was described by a linear function, whereas the relationship with the duration of weed-infested period showed a better fit with a logistic function. To maintain 90% of maximum seed yield, the maximum time allowed to let weeds grow after the crop emergence varied over locations from 4.8 to 5.8 wk. The same level could be achieved if the crop is kept free of weeds from its emergence until 12.1 to 14.1 wk; while the time when the same amount of yield would be achieved under both approaches varied from 7.7 to 9.3 wk after crop emergence. For straw yield, the time to get 90% of the maximum yield could vary over location from a maximum of 4.5 to 8.0 wk under weed-infestation and from at least 11.5 to 13.5 wk when weed-free. The time to achieve the same amount of straw under two systems of competition varied from 6.5 to 9.9 wk after crop emergence. One of the four experiments showed a longer critical period than the others for seed and straw yields.