The airline passenger demand forecasting is quite challenging since the demand data includes both seasonality and growth trend over several years. Traditional forecasting methodology uses the known parameters of a forecasting method based on the product/service life cycle phase. This paper aims to develop a methodology for finding optimum parameters for accurate demand forecasting of airline passengers using the latest dataset of the US department of transportation. The proposed methodology first analyzes demand data, cleanses data, identifies pattern, removes seasonality, uses a forecasting method (trend adjusted exponential smoothing, i.e., TAF), finds optimum parameters (set of smoothing constant α and trending constant β) using nonlinear programming (NLP) to minimize the error measures: Mean Absolute Deviation (MAD), Mean Square Error (MSE), and Mean Absolute Percentage Error (MAPE) to find demand forecasts. Furthermore, multicriteria NLP is used to minimize MAD, MSE and MAPE to a satisfactory level to improve α and β for demand forecasting. In this paper, a specific route of an airline (Air Canada, SFO to YYZ) is selected, and the methodology produced demand forecasts that are less than 0.1% close to those of the minimum MAD, MSE, MAPE, thereby improving on all these three measure at the same time when compared to the actual demand of the year 2019. Later, the same methodology is applied to find the forecasts for the year 2024.