The paradigm that meiotic recombination and chiasmata have the same basis has been challenged, primarily for plants. High resolution genetic mapping frequently results in maps with lengths far exceeding those based on chiasma counts. In addition, recombination between specific homoeologous chromosomes derived from interspecific hybrids is sometimes much higher than can be explained by meiotic chiasma frequencies. However, almost the entire discrepancy disappears when proper care is taken of map inflation resulting from the shortcomings of the mapping algorithm and classification errors, the use of dissimilar material, and the difficulty of accurately counting chiasmata. Still, some exchanges, especially of short interstitial segments, cannot readily be explained by normal meiotic behaviour. Aberrant meiotic processes involving segment replacement or insertion can probably be excluded. Some cases of unusual recombination are somatic, possibly premeiotic exchange. For other cases, local relaxation of chiasma interference caused by small interruptions of homology disturbing synaptonemal complex formation is proposed as the cause. It would be accompanied by a preference for compensating exchanges (negative chromatid interference) resulting from asymmetry of the pairing chromatid pairs, so that one side of each pair preferentially participates in pairing. Over longer distances, the pairing face may switch, causing the normal random chromatid participation in double exchanges and the relatively low frequency of short interstitial exchanges. Key words : recombination frequency, map length, chiasmata, discrepancy, chromatid interference.