In this correspondence, fractional negative-order moments are used to estimate the K-clutter plus noise parameters. Combining the fractional positive and negative moments, we aim to improve the accuracy of the estimation outcomes. This is achieved through the reduction of the effect of the two first intensity moments involved in the equation of the unknown parameters. For instance, the proposed fractional positive-and negative-order moments' estimator, given in terms of two confluent hypergeometric functions, is solved numerically to yield the shape parameter. Using single pulse and multiple pulses in various clutter plus noise scenarios, we show, via both simulated and IPIX real data, a comparison of the new estimator with the estimators based on the higher-order moments, the fractional positive-order moments, the [zlog(z)], and the constrained maximum likelihood.