We study mesoscopic disorder fluctuations in an anisotropic gap superconductor, which lead to the spatial variations of the local pairing temperature and formation of superconducting islands above the mean-field transition. We derive the probability distribution function of the pairing temperatures and superconducting gaps. It is shown that above the mean-field transition, a disordered BCS superconductor with an unusual pairing symmetry is described by a network of superconducting islands and metallic regions with a strongly suppressed density of states due to superconducting fluctuations. We argue that the phenomena associated with mesoscopic disorder fluctuations may also be relevant to the high-temperature superconductors, in particular, to recent STM experiments, where gap inhomogeneities have been explicitly observed. It is suggested that the gap fluctuations in the pseudogap phase should be directly related to the corresponding fluctuations of the pairing temperature. PACS numbers: 74.40.+k, 74.81.Bd, 74.20.De Understanding the phase diagram and the properties of the high-temperature and other unconventional superconductors has been among the most complex problems of modern condensed matter physics. Most current theoretical approaches to the problem concentrate on strong correlation physics and usually assume that the effects of disorder are unimportant. However, there exist a number of recent experimental works, in particular STM studies of the high-T c cuprates, 1,2,3,4,5,6,7 which provide a tentative indication that at least in some materials disorder plays an important role in the local formation of the superconducting gap. In particular, Gomes et al. 2 have studied the local development of the gap as a function of temperature in Bi 2 Sr 2 CaCu 2 O 8+δ above the superconducting transition and up to a pseudogap temperature, where the gap inhomogeneities cease to exist. An important result of this experiment is that the real-space gap map observed was static and reproducible. This strongly suggests that the inhomogeneous gap formation is unlikely to be a phase-separation or superconducting fluctuation effect, but is due to some kind of disorder in the system.Motivated by these experiments, we theoretically consider a disordered superconductor with an unusual pairing symmetry (e.g., a d-wave superconductor) and study mesoscopic variations of the local pairing temperature. We point out that the existence of the Griffiths-type 8,9,10 phase in the superconducting phase diagram is specific to an anisotropic gap superconductor and should not occur in the conventional swave systems (due to Anderson theorem), unless they are extremely dirty or time-reversal symmetry is broken. 11,12,13 An important observation is that if the pairing gap is anisotropic, the Anderson theorem breaks down and the superconducting pairing temperature, T p , is suppressed by disorder even if time-reversal symmetry is preserved (here and below we make a distinction between the pairing temperature, T p , and the superconducting trans...