BACKGROUND: It is generally accepted that a UVA-induced erythema is difficult to detect except in the most sensitive individuals. OBJECTIVE AND METHODS: As UVA effects on human skin and skin cells have been shown to depend strongly on anatomical body sites, UVA I, UVA I + II and solar simulator radiations were compared in their ability to induce erythema and melanin pigmentation responses in individuals with skin types I-IV on both previously sun-exposed (arms, forearms, thighs) and nonexposed body sites (buttocks). RESULTS: Erythema induction by UVA I on previously nonexposed skin sites followed a dose response in all skin types which was contrary to the absence of erythema induction seen on previously sun-exposed sites. Melanin expression followed a dose and skin type response and was shown to be more enhanced in previously exposed skin and in skin types III and IV. In contrast, UVA I + II induced erythema on nonexposed skin areas and to a lesser extent on frequently sun-exposed skin. Melanin production by UVA I + II was similar to that seen with UVA I alone in individuals of skin types II and III. Solar simulator radiation was very efficient in erythema induction regardless of previous sun exposure of skin. CONCLUSIONS: We have found that contrary to the widespread opinion that UVA and in particular UVA I could not induce a significant erythema, this waveband is capable of measurable erythema induction on skin nonexposed to sunlight. The diminished erythema induction by UVA I on chronically sun-exposed skin suggests the possibility of a defense mechanism against UVA-induced damage in this tissue