We propose a low complexity complex valued Sphere Decoding (CV-SD) algorithm, referred to as Circular Sphere Decoding (CSD) which is applicable to multiple-input multiple-output (MIMO) systems with arbitrary two dimensional (2D) constellations. CSD provides a new constraint test. This constraint test is carefully designed so that the element-wise dependency is removed in the metric computation for the test. As a result, the constraint test becomes simple to perform without restriction on its constellation structure. By additionally employing this simple test as a prescreening test, CSD reduces the complexity of the CV-SD search. We show that the complexity reduction is significant while its maximum-likelihood (ML) performance is not compromised. We also provide a powerful tool to estimate the pruning capacity of any particular search tree. Using this tool, we propose the Predict-And-Change strategy which leads to a further complexity reduction in CSD. Extension of the proposed methods to soft output SD is also presented.