The author begins by tracing the history of the concept of empathy in psychoanalysis, noting that, largely through the influence of Kohut, it began to feature prominently in the literature from the late 1950s on and has since tended, wrongly in his opinion, to be regarded as an all-purpose instrument to be deployed at will. What is often described in theoretical contributions as empathy should in the author's view more properly be called concordance. On the clinical level, the idea that the analyst must deliberately seek to empathise with the patient is stated to have gained currency, but the author argues that such an attempt to achieve empathy by force can lead only to 'empathism', which is a dogmatic, hyperconcordant attitude whereby the inexperienced analyst in particular thinks he can control the process better. Clinical material is presented to show how some patients set out to induce 'empathism' in the analyst for defensive reasons and how the analyst's concordance, until analysed, may lead to an impasse. The author stresses that genuine empathy is a state of complementary conscious-preconscious contact based on separateness and sharing; covering not only the patient's ego-syntonic subjectivity but also his defensive ego and split-off parts, its achievement requires prolonged hard work on the countertransference and a capacity for contact with the analyst's own primitive aspects. The paper ends with a consideration of the possible obstacles to empathic contact.