Two experiments of Pavlovian conditioning with rabbits evaluated the effects of initiating or continuing a conditioned stimulus (CS) after a paraorbital unconditioned stimulus (US). In Experiment 1, backward pairings, in which a CS carne on after the US, produced a CS that appeared inhibitory on a measure of eyeblink conditioning but excitatory on a potentiated-startle measure of conditioned fear. In Experiment 2, extending the duration of a CS that carne on prior to the US,so that it continued after the US, decreased eyeblink conditioned responses, whereas it increased conditioned fear. The data from the two experiments confirm and extend those of Tait and Saladin (1986), supporting the suppositions of AESOP (Wagner & Brandon, 1989) that conditioned eyeblink and conditioned fear can be dissociated under various temporal relationships between the CS and US. Tait and Saladin (1986) reported that "backward" training, where an aversive unconditioned stimulus (US) was followed, rather than preceded, by a conditioned stimulus (CS), produced a CS that appeared concurrently excitatory and inhibitory, depending on the measure of associative learning employed. After 65 conditioning trials, where a 1,000-msec tone CS was administered 500 msec after a 1OO-msec paraorbital shock US, the tone appeared excitatory as indicated by its acquired ability to suppress drinking behavior, whereas it appeared inhibitory as evidenced by its retardation in subsequent acquisition ofeyeblink conditioned responses (CRs). Wagner and Brandon (1989) pointed to these data as providing especially clear evidence that CS-US pairings can produce an association between the CS and the emotive qualities of the US (as presumably reflected in common measures of conditioned fear) that is separable from the association between the CS and the remaining sensory-perceptual attributes ofthe US (as presumably seen in discrete CRs, such as the eyeblink). The separability is such that, under the conditioning parameters employed by Tait and Saladin, one association may be excitatory while the other is inhibitory.The data reported by Tait and Saladin (1986) might have been anticipated on the basis of the separate literatures on fear conditioning and on eyeblink conditioning concerned with the effects of backward US-CS pairings. Studies offear conditioning that have employed relatively short US-CS intervals have repeatedly reported an excitatory associative consequence