Hand eczema (HE) is a common inflammatory dermatosis that causes significant patient morbidity. Symptoms such as pain, itch and localized swelling contribute to disruption of activities of daily living. The aetiology is often multifactorial. Previous studies comparing psoralen–ultraviolet A (PUVA) vs. narrowband ultraviolet B (NB-UVB) for HE have been small, nonrandomized and retrospective, and are therefore unsuitable to use as pilot data for a larger study. We performed a randomized, observer-blinded pilot study of PUVA vs. NB-UVB for the treatment of hand eczema. Hand eczema severity was assessed using Physician’s Global Assessment (PGA) and a validated scoring system called the modified Total Lesion and Symptom Score (mTLSS). The primary outcome measure was patients achieving a ‘clear’ or ‘almost clear’ treatment response at 12 weeks. Secondary outcome measures included assessment of mTLSS, Dermatology Life Quality index (DLQI) and the EuroQol health outcome score (EQ-5D). Sixty patients aged over 18 years with HE unresponsive to clobetasol propionate were randomized to receive immersion PUVA (n = 30) or NB-UVB (n = 30) twice weekly for 12 weeks with 4- weekly assessments. The main inclusion criterion was palmar eczema unresponsive to topical treatments. In the PUVA arm, 23 patients had a PGA assessment at 12 weeks. Of these, four were ‘clear’, and eight were ‘almost clear’ (response rate 40%). In the NB-UVB group, out of 20 patients assessed at 12 weeks, two were ‘clear’ and four ‘almost clear’ (response rate 20%). For PUVA, the median (range) baseline mTLSS scores for the left and right hands, respectively (n = 30) were 8.5 (0–16) and 8 (3–15), while at 12 weeks (n = 23) the scores were 3 (0–13) and 3 (0–14). For NB-UVB, the median baseline mTLSS results for the left and right hands (n = 30) were 7 (0–16) and 8.5 (1–15), and at 12 weeks (n = 20) the scores were 5 (0–11) and 4.5 (0–11). The median DLQI scores in the PUVA and NB-UVB groups changed from 9.5 (5–27) and 10 (2–27) at baseline to 2 (0–21) and 4 (0–22) at 12 weeks, respectively. There were no treatment-related moderate-to-severe adverse events reported in patients randomized to PUVA, and nine moderate-to-severe adverse events (mainly erythema) in the group randomized to NB-UVB. There were no treatment-related serious adverse events. This pilot study shows the superiority of PUVA, but also a response to NB-UVB. Further optimization of NB-UVB dosimetry may improve outcomes and may be a worthwhile option for patients wanting to avoid PUVA.