In backward visual masking, it is common to find that the mask has its biggest effect when it follows the target by several tens of milliseconds. Research in the 1960s and 1970s suggested that masking effects were best characterized by the stimulus onset asynchrony (SOA) between the target and mask. In particular, one claim has been that the SOA for which masking is optimal remains fixed, even as target and mask durations varied. Experimental evidence supported this claim, and it was accepted as an SOA law. However, recent modeling (Francis, 1997) and experimental studies (Macknik and Livingstone, 1998) argued for new ISI (interstimulus interval) and STA (stimulus termination asynchrony) laws, respectively. This paper reports a mathematical analysis and experimental tests of the laws. The mathematical analysis demonstrates unsuspected relationships between the laws. The experiments test the predictions of the SOA, ISI, and STA laws. The data favor the ISI law over the SOA and the STA laws.