Bit-loaded orthogonal frequency-division multiplexing (OFDM) with convolutional coding is a powerful technique for transmission over quasi-static frequency-selective fading channels. Further enhancements in data rate are achieved by combining loaded OFDM with multiple-input-multiple-output (MIMO) transmission. Motivated by the lack of appropriate error-rate analysis techniques for this popular type of transmission system, in this paper, we develop a novel analytical method for bit-error-rate (BER) and frame-error-rate (FER) estimation of bit-loaded coded OFDM and MIMO-OFDM systems using singular value decomposition (SVD), operating over frequency-selective quasi-static channels with nonideal interleaving. Then, we introduce three different applications of the proposed analysis. First, we compare the performance of several OFDM bit-loading schemes and propose a hybrid loading scheme that selects the best loading for each channel realization from a number of candidates. Second, we introduce three adaptive interleaving schemes: 1) selecting the best interleaver from a number of predefined interleavers; 2) a novel adaptive bit-interleaving algorithm based on the pairwise error probability; and 3) a spatial interleaving scheme for MIMO-OFDM-SVD systems with separate information sources. Third, we introduce an adaptive coded-modulation algorithm by using our BER and FER estimation techniques.