Adult male rats were castrated and implanted sc with testosterone (T)-filled Silastic capsules of different sizes which produce stable plasma T levels proportional to capsule length. T and LH levels were determined in blood samples obtained at 3-day intervals for 21 days. Plasma T values greater than 1.8 ng/ml suppressed plasma LH to levels below the intact range, while a plasma T of less than 0.5 ng/ml yeilded no significant inhibition of the post-castration LH rise. Animals with a plasma T of about 0.7 ng/ml, however, showed intact, castrate, or immediate LH levels. The feedback-effective range for all animals was between 0.5 ngT/mg (castrate-like LH response) and 1.8 ng/ml (LH at or below intact levels), but individual rats seemed to have a more restricted feedback range than that indicated by these group data. To test this hypothesis further, rats were implanted at the time of castration with five 5-mm T-filled capsules. One capsule was removed each week, producing step function drops in plasma T, and allowing examination of the LH responses of individual animals. Plasma t levels greater than 1.6 ng/ml maintained LH values at or below intact levels. However, when plasma T dropped to 1.1 ng/ml, some rats showed castration-like LH increase and others did not respond until after a further drop to 0.6 ng/ml. It was concluded that the feedback-effective range of plasma T is very restricted and differs in individual rats.