The author considers parallel iterative methods for systems of linear equations \(Au=d\), \(A=A^*>0\). If \(A=B_ k-C_ k\), \(k=1,...,K\) are given splittings of the matrix A then the iterative methods can be written in the form \(u^{n+1}=\sum_{k}D_ kB_ k^{-1}C_ ku^ n+\sum_{k}D_ kB_ k^{-1}d\) where \(D_ k\geq 0\) are diagonal matrices and \(\sum_{k}D_ k=I\). Convergence of the methods is analyzed and special attention is paid to variants of the successive overrelaxation method for differential systems. Numerical experiments indicate the effectiveness of the methods.