EPA and DHA are often used in veterinary medicine due to their beneficial effects for several medical conditions such as osteoarthritis. EPA and DHA are administered to dogs through different matrices. The aim of the present study was to determine the effects on the plasma levels in dogs caused by various matrices for EPA and DHA administration. In this study, three different n-3 PUFA formulations were used: soft chew tablet (CCx); liquid fish oil (LFO); and enriched kibbles (EK). The formulations were administered single-dose and compared in a randomised, cross-over designed study with a 1-week wash-out period. Several variables were observed after the administration of these formulations in thirteen dogs: the NEFA plasma concentration, the AUC for 1 d (AUC0–24 h), and maximum plasma concentration for both EPA and DHA. All plasma fatty acid levels reached baseline levels within 72 h. CCx (median = 2·987) had a significantly lower AUC0–24 h for EPA compared with LFO (median = 5·647, P = 0·043) and EK (median = 5·119, P = 0·032) (F2,22 = 4·637, P = 0·021). CCx (median = 2·471) AUC0–24 h for DHA was significantly lower compared with LFO (median = 4·837, Z = −2·56, P = 0·011) and EK (median = 4·413, Z = −2·59, P = 0·01). EPA and DHA plasma levels were affected by matrix, as with the CCx, the AUC0–24 h of EPA and DHA were both lower compared with LFO and EK. The effect of matrix on bioavailability is important for product development as well as for clinical trials studying effects of EPA and DHA.