We derive an analytical formula for the covariance Cov(A, B) of two smooth linear statistics A = i a(λi) and B = i b(λi) to leading order for N → ∞, where {λi} are the N real eigenvalues of a general one-cut random-matrix model with Dyson index β. The formula, carrying the universal 1/β prefactor, depends on the random-matrix ensemble only through the edge points [λ−, λ+] of the limiting spectral density. For A = B, we recover in some special cases the classical variance formulas by Beenakker and Dyson-Mehta, clarifying the respective ranges of applicability. Some choices of a(x) and b(x) lead to a striking decorrelation of the corresponding linear statistics. We provide two applications -the joint statistics of conductance and shot noise in ideal chaotic cavities, and some new fluctuation relations for traces of powers of random matrices.Introduction -The discovery of the phenomenon of universal conductance fluctuations (UCF) in disordered metallic samples, pioneered by Altshuler [1] and Lee and Stone [2] has had a profound impact on our current understanding of the mechanisms of quantum transport at low temperatures and voltage. There are two aspects of this universality, i.) the variance of the conductance is of order (e 2 /h) 2 , independent of sample size or disorder strength, and ii.) this variance decreases by precisely a factor of two if time-reversal symmetry is broken by a magnetic field. Both features, observed in several experiments and numerical simulations (see [3] for a review), naturally emerge from a random-matrix theoretical formulation of the electronic transport problem [4,5]. The phenomenon of UCF is just, however, one of the very many incarnations of a more general and intriguing property of sums of strongly correlated random variables.