Response spectra are commonly used for the estimation of the largest peak response of a linear structural system in a seismic environment. Traditionally, this has been done through the use of appropriate modal combination rules in case of multi-degree-of-freedom systems. While these methods do not consider uncertainty in response due to phasing in seismic waves, those also do not go beyond estimating the largest peak response, and have natural limitation of being applied accurately to only a few types of structural systems. This paper considers a review of alternative methods which have been developed since mid-1970’s to give probabilistic estimates of response peaks, while continuing to use the information available through response spectra. These methods have the convenience of being applied in a variety of situations, do not usually suffer from the inaccuracies associated with the use of modal combination rules, and present state-of-the-art methodology in linear seismic response analysis. The limitations of various formulations proposed under these methods are identified, and future directions of required work are suggested.