Examples of wideband spectrum sensing techniques for cognitive radio are provided. The techniques permit undersampling (e.g., sub-Nyquist sampling) the spectrum band, which permits use of more cost effective electronic components (e.g., lower frequency analog-to-digital converters (ADC)). The techniques permit determination of whether there are any (or sufficient) sub-channels that exist in the band to use for transmission, so that only a portion of less than the entire spectrum band is sampled. Detection decisions can be based at least partly on a pair of performance metrics more appropriate for wideband spectrum sensing. These performance metrics are the probability of insufficient spectrum opportunities (P.sub.ISO) and the probability of excessive interference opportunities (P.sub.EIO).