We studied the limitations to early diagnosis in amyotrophic lateral sclerosis (ALS). The diagnostic process was assessed in 120 consecutive patients, including onset, interval to diagnosis, investigations, specialist assessment and pre-diagnostic management. Times from onset to first consultation (T1), second consultation (T2) and diagnosis (TD) were considered. Predictors of diagnostic delay were determined by multivariate logistic regression, adjusted for gender, age, clinical manifestations, and specialism of the first and second consultants. There were 101 consecutive ALS patients with complete datasets (69% men; median age at diagnosis 61.5 years). The mean TD and median TD were respectively 10.1 and 9.5 months. In 55%, the first consultant was a general practitioner (GP), in 16% a neurologist and in 14% an orthopedist. The diagnosis of ALS was made by non-neurologists in 9 patients. The odds of delayed diagnosis (≥ 12 months) were higher (1.56; 0.19–12.56) in younger patients (≤ 45 years) (p < 0.05). Female gender (0.56; 0.29–1.70) and bulbar-onset (0.56; 0.29–1.70) were independently associated with earlier diagnosis (p < 0.05). Assessment by a neurologist at the first (0.32; 0.19–2.46) or second consultation (0.87; 0.21–1.21) was associated with a shorter diagnosis time (< 12 months) (p < 0.05). We conclude that diagnostic delay mainly resulted from delayed referral from non-neurologist physicians to a neurologist. Moreover, incomplete neurophysiological investigation had a relevant impact.