Optimal control of Markov-modulated SDEs with random coefficients is considered. The linear quadratic case is solved in terms of the associated backward stochastic Riccati equation.