This paper deals with the design of a reduced-order H-infinity filter for a stochastic bilinear system with a prescribed X., norm criterion. The considered system is bilinear in control and with multiplicative noises in the dynamics and in the measurement equations. The problem is transformed into the search of a unique gain matrix by using Sylvester-like constraints. The approach is based on the resolution of LMI and is then easily implementable. (C) 2009 Elsevier Ltd. All rights reserved.