In this paper, we present a new tractable expression for average probability of Energy Detection (ED) in generalized-K (K G) fading channel. ED has been proposed for spectrum sensing in cognitive radios and various wireless applications. A closed form solution for modeling ED under shadow-fading conditions using lognormal distribution is complicated and is the main drawback. In K G fading model, shadowing is modeled by Gamma probability density function (PDF) and Nakagami-m distribution is used to represent the small-scale random variations of the received signal envelope. By applying the alternative series representation of Marcum-Q function, We obtain a closed form expression for ED based average detection probability over K G fading. The derived expressions are validated, and analytical results are verified by numerical computations and simulations for scenarios of practical interest Further, these results are considered for performance analysis and improvement of sensing by varying parameters of K G fading channel under light and moderate shadowing.