Patients from a randomised trial on resurfacing hip arthroplasty (RHA) (n = 36, 19 males; median age 57 years, 24 to 65) comparing a conventional 28 mm metal-on-metal total hip arthroplasty (MoM THA) (n = 28, 17 males; median age 59 years, 37 to 65) and a matched control group of asymptomatic patients with a 32 mm ceramic-on-polyethylene (CoP) THA (n = 33, 18 males; median age 63 years, 38 to 71) were cross-sectionally screened with metal artefact reducing sequence-MRI (MARS-MRI) for pseudotumour formation at a median of 55 months (23 to 72) post-operatively. MRIs were scored by consensus according to three different classification systems for pseudotumour formation. Clinical scores were available for all patients and metal ion levels for MoM bearing patients. Periprosthetic lesions with a median volume of 16 mL (1.5 to 35.9) were diagnosed in six patients in the RHA group (17%), one in the MoM THA group (4%) and six in the CoP group (18%). The classification systems revealed no clear differences between the groups. Solid lesions (n = 3) were exclusively encountered in the RHA group. Two patients in the RHA group and one in the MoM THA group underwent a revision for pseudotumour formation. There was no statistically significant relationship between clinical scoring, metal ion levels and periprosthetic lesions in any of the groups. Periprosthetic fluid collections are seen on MARS-MRI after conventional CoP THA and RHA and may reflect a soft-tissue collection or effusion. Currently available MRI classification systems seem to score these collections as pseudotumours, causing an-overestimatation of the incidence of pseudotumours.