SUMMARY (1) Analysis of published studies generally indicates that ecological succession can be considered as a non-random process. (2) Two examples are discussed in detail, termite succession on baitwood blocks in Ghana (Usher 1975) and predator-prey dynamics of mites in a complex universe of oranges (Huffaker 1958), and both indicate that succession is a non-stationary Markovian process. (3) The discussion considers that complex non-random or Markovian processes are likely to characterize ecological successions, the transition probability matrix elements not being constant but being functions either of the abundance, or of the rate of change of abundance, of a recipient class. (4) Tests of various hypotheses, using x2 criteria, are given.