The role of multiplicative colored noise on the photon statistics of the dye-laser output is investigated. This model explains consistently the recent experimental results by Short, Mandel, and Roy. PACS numbers: 42.55.Mv, 42.50. + q The physics of macroscopic systems can often be described in terms of a few collective variables whose equations of motion are obtained by the elimination of microscopic variables and are, in general, highly nonlinear. If the system is in stable equilibrium, the statistical nature of the processes arising due to the microscopic dynamics can be neglected. However, in the case where the system approaches the limit of stability, the fluctuations due to microscopic variables play a very crucial role in governing the evolution of the system and hence must be incorporated into the equation of motion for macroscopic variables. The fundamental importance of fluctuations has been realized in a large number of systems, e.g., the laser, autocatalytic reactions, hydrodynamic and current instabilities, and equilibrium phase transitions for which the deterministic macroscopic description is found to be inadequate, especially near the threshold regime.There is currently intense activity in analyzing the effect of fluctuations on the dynamics of nonlinear processes. 1 These fluctuations, appearing as additive and/or multiplicative 2 noise terms in the equations of motion, are usually modeled at 6-correlated Gaussian processes (or white noise). This assumption, although convenient for mathematical analysis, is somewhat unrealistic as the fluctuations arising due to the microscopic dynamics would have a finite (nonzero) correlation time, (Such fluctuations are commonly referred to as colored noise.) Only in certain regions of parameter space, where all other retevent times in the problem are much longer than the correlation time of the fluctuations, would the whitenoise approximation be valid while in other regions discrepancies between white-noise theory and experiment would become noticeably large.The problem of photon statistics of the dyelaser output falls into this category as the equation of motion for the electric field e(t) is nonlinear and the colored noise is important. Re-cent experiments 3 * 4 have demonstrated some very interesting statistical properties of e(t); e.g., the relative intensity fluctuations, defined as ((A/) 2 )/(/) 2 , tend to increase indefinitely as the laser is weakly excited (}-0). Such a behavior, inexplicable in terms of additive noise, is suspected to arise due to fluctuations in the pump parameter that appear multiplicatively in the equation of motion for e(t). The theory proposed by Graham, Hohnerbach, and Schenzle,5 where the noise was modeled as a Gaussian 6-correlated (white-noise) process, although successful in explaining some of the experimental results of Ref. 4, fails to explain results in another parameter regime as has been demonstrated by Short, Mandel, and Roy 4 in a recent publication.It is the purpose of this Letter to explain the experimental ...