Common haptic devices are designed to effectively provide kinaesthetic and/or cutaneous discriminative inputs to the users by modulating some physical parameters. However, in addition to this behavior, haptic stimuli were proven to convey also affective inputs to the brain. Nevertheless, such affective properties of touch are often disregarded in the design (and consequent validation) of haptic displays. In this paper we present some preliminary experimental evidences about how emotional feelings, intrinsically present while interacting with tactile displays, can be assessed. We propose a methodology based on a bidimensional model of elicited emotions evaluated by means of simple psychometric tests and statistical inference. Specifically, affective dimensions are expressed in terms of arousal and valence, which are quantified through two simple one-question psychometric tests, whereas statistical inference is based on rank-based non-parametric tests. In this work we consider two types of haptic systems: (i) a softness display, FYD-2, which was designed to convey purely discriminative softness haptic stimuli and (ii) a system designed to convey affective caress-like stimuli (by regulating the velocity and the strength of the “caress”) on the user forearm. Gender differences were also considered. In both devices, the affective component clearly depends on the stimuli and it is gender-related. Finally, we discuss how such outcomes might be profitably used to guide the design and the usage of haptic devices, in order to take into account also the emotional component, thus improving system performance.