Background Universal newborn hearing screening (UNHS) targets moderate or greater hearing loss. However, UNHS also frequently detects children with mild loss that results in many receiving early treatment. The benefits of this approach are not yet established. We aimed to (i) compare language and psychosocial outcomes between four hearing loss detection systems for children aged 5–8 years with congenital mild–moderate hearing loss; (ii) determine whether age of detection predicts outcomes; and (iii) compare outcomes between children identified via well‐established UNHS and the general population. Methods Linear regression adjusted for potential confounding factors was used throughout. Via a quasi‐experimental design, language and psychosocial outcomes were compared across four population‐based Australian systems of hearing loss detection: opportunistic detection, born 1991–1993, n = 50; universal risk factor referral, born 2003–2005, n = 34; newly established UNHS, born 2003–2005, n = 41; and well‐established UNHS, born 2007–2010, n = 21. In pooled analyses, we examined whether age of detection predicted outcomes. Outcomes were similarly compared between the current well‐established UNHS system and typically developing children in the Early Language in Victoria Study, born 2003, n = 1217. Results Age at diagnosis and hearing aid fitting fell steadily across the four systems. For moderate losses, mean expressive language (P for trend .05) and receptive vocabulary (P for trend .06) improved across the four systems, but benefit was not obvious for mild losses. In pooled analyses, diagnosis before age six months predicted better language outcomes for moderate losses. Children with mild–moderate losses exposed to well‐established UNHS continue to experience expressive language scores well below children in the general population (adjusted mean difference −8.9 points, 95% CI −14.7 to −3.1). Conclusions Treatment arising from UNHS appears to be clearly benefitting children with moderate hearing losses. However, rigorous trials are needed to quantify benefits, versus costs and potential harms, of early aiding of children with mild losses.