Background Medical nutritional therapy is an important component of type 1 diabetes (T1D) care in children and carbohydrate counting is one such method. We aimed to evaluate the knowledge of carbohydrate counting among parents of children with T1D from Sri Lanka and study its association with the child's glycaemic control. Methods A descriptive cross-sectional study was conducted among parents of children with T1D. HbA1c measurement was used to assess glycaemic control. Knowledge of parent regarding carbohydrate counting was assessed based on a 24-hour dietary recall. Carbohydrate counting knowledge was defined using ratio of carbohydrate content estimated by parents to actual carbohydrate content calculated by researchers (Total, Breakfast, Lunch, Dinner, and Snacks). Ratios obtained were also divided into three groups, underestimation (<0.9), accurate estimation (0.9-1.1), and overestimation (>1.1). A multivariate regression analysis was performed to determine contribution of carbohydrate counting accuracy to glycaemic control (HbA1c). Results Sample size was 181 and mean age of the parents was 38.8±5.9 years. Mean duration of diabetes in the children was 3.7±2.6 years and mean HbA1c level was 8.3±0.9%. On average, parents estimates of carbohydrate count for the total meal were 0.88±0.27 (88%) (range 0.38-1.47) of the actual carbohydrate count. Only 30.5% (n=55) of parents were grouped in the “accurate” estimation category for the total carbohydrate count. Parents of children with diabetes for ≤3 years estimated total carbohydrate count more accurately than the counterparts (p<0.05). Mean HbA1c value of those who “underestimated” was significantly higher than those with “accurate” estimation. In the multivariate analysis accuracy of carbohydrate estimation was associated with a lower HbA1c (β = −0.36; p=0.03). Conclusions Overall knowledge of carbohydrate counting among parents was inadequate. Better knowledge was associated with improved glycaemic control in children and lower incidence of hypoglycaemic episodes. An inverse association was observed between knowledge and duration of diabetes.