In this paper, we address the problem of identifying fixed order stable SISO systems from time and frequency domain data. Given measurements of time and frequency response corrupted by process and measurement noise, we aim at finding a fixed order stable plant whose response matches the data collected (within the noise bounds) and whose H∞ norm is below a prescribed level. It is shown the problem can be solved by finding a point in a set defined by polynomial inequalities and the sparse structure of the polynomials is exploited to develop an efficient system identification algorithm. Further computational improvements are obtained by reformulating the problem as a rank constrained one and using efficient convex relaxations of rank minimization. Numerical examples are provided to illustrate the efficiency of the proposed algorithms.