PurposeWeight changes occur throughout the cancer trajectory. Most research has focused on changes during or after treatment, so clinical significance of change at diagnosis remains unclear. This study aimed to determine prevalence, predictors and prognostic significance of weight changes at diagnosis in outpatients with solid tumours presenting to a tertiary academic medical centre.MethodsA retrospective study of the electronic medical record was conducted (n=6477). Those with weight recorded within 6months of cancer diagnosis (pre-diagnosis, T-0) and 2 subsequent weights (diagnosis, T-1; final visit, T-2) were identified (n=4258). Percentage weight change was categorised into four bands (0.1-2.4%; 2.5-5%; 5.01-9.9%; 10%) for gain and loss. A stable category was also included.ResultsMean age is 6112.5years. Common tumour sites: breast (17%; n=725), prostate (16%; n=664), lung (14%; n=599). 15% (n=652) had metastatic disease at T-1. 98% (n=4159) had weight change at T-1. Head & neck and upper gastrointestinal cancers were significantly associated with weight loss (p<0.001). Worst survival occurred with 10% weight gain or 10% weight loss. Overweight or obese body mass index with any percentage weight change band was associated with better overall survival.Conclusions Most had evidence of clinically significant weight changes at diagnosis. Weight loss at diagnosis was associated with a higher risk of further weight loss. A detailed weight history at cancer diagnosis is essential to identify and intervene for those most at risk of weight change-related early mortality.