Algebraic space-time coding allows for reliable data exchange across fading multiple-input multiple-output channels. A powerful technique for decoding space-time codes in Maximum-Likelihood (ML) decoding, but well-performing and widely-used codes such as the Golden code often suffer from high ML-decoding complexity. In this article, a recursive algorithm for decoding general algebraic space-time codes of arbitrary dimension is proposed, which reduces the worst-case decoding complexity from $O(|S|^{n^2})$ to $O(|S|^n)$.