1. Sixty-seven afferents from the finger extensor muscles were consecutively recorded by microneurography. 2. The units were classified as primary or secondary muscle spindle afferents or Golgi tendon organ afferents on the basis of their responses to ramp-and-hold stretches, sinusoidals superimposed on ramp-and-hold stretches, maximal twitch contractions and isometric contractions and relaxations. 3. The muscle was repeatedly stretched and then either kept short or long for a few seconds followed by a slow ramp stretch. The responses of the muscle afferents to the slow stretch were compared under the two conditions. 4. Thirty out of thirty-eight units classified as primary spindle afferents and four out of eleven units classified as secondary afferents showed an enhanced response to the slow ramp when the muscle had been kept short compared to the response when the muscle had been kept long. 5. None of the eighteen Golgi tendon organ afferents showed any difference in this respect. 6. It is concluded that stretch sensitization does occur in human muscle spindles and, when present, constitutes firm evidence of the afferent originating from a muscle spindle rather than a Golgi tendon organ. In addition, due to differences in the response characteristics of primaries and secondaries, the test may aid in separating muscle spindle primary afferents from secondary afferents.