This paper utilizes iterative covariance steering to handle the atmospheric entry guidance problem with stochastic atmosphere density. The stochastic entry guidance is formulated as a nonlinear covariance steering problem, where atmosphere density is modeled by Gaussian Random Field, and the path constraints are considered as chance constraints. The problem is then approximated by a convex subproblem via successive convexification of uncertainty propagation and the relaxation of chance constraints. After considering an adaptive trust region, the overall problem is solved by iteratively optimizing the subproblem until the solution converges. Numerical simulations for Mars entry show that the solution is converged within five iterations, and the distribution of final states and the path constraints meet the requirements.