Built-in Self-Test (BIST) is essential, particularly for radio frequency (RF) devices where off-chip RF signal analysis is costly or in some cases, infeasible. Two major problems have made RF BIST elusive. First, process variations make the BIST circuit behavior hard to predict, limiting accuracy of measurements. Second, the overhead, particularly in terms of performance degradation, make RF BIST undesirable. In this paper, we address these two issues for RF BIST gain measurement. First, we show that by setting up relative gain measurements and carefully crafting the BIST methodology and the matching BIST circuit, the effect of process variations on measurement accuracy can be suppressed. Second, by codesigning the BIST circuit together with the device under test (DUT), performance impact can be eliminated or significantly reduced. To demonstrate the proposed approach, we design a low noise amplifier (LNA) as the DUT together with the BIST circuit. We also design a stand-alone LNA with the same specifications and manufacture these two circuits on the same die. We show that the LNA gain can be determined very accurately, using only DC measurements, and the performance impact of the BIST circuit is negligible.