This paper proposes using a quasi-random sequence for estimating the mixed multinomial logit (MMNL) model. The MMNL model is a flexible discrete choice model that accommodates general patterns of competitiveness and heterogeneity in individual sensitivity to exogenous variables. Traditional estimation methods, such as pseudo-random maximum simulated likelihood, evaluate multi-dimensional integrals in the log-likelihood function by averaging integrand values at pseudo-random points. The authors suggest an alternative quasi-random maximum simulated likelihood method using uniformly distributed sequences, which provides better accuracy with fewer draws and computational time. Numerical experiments on intercity travel mode choice show that the quasi-random method outperforms the pseudo-random method. This has potential to significantly influence the practical use of the MMNL model, especially given its flexibility in accommodating complex behavioral structures. The paper discusses three numerical integration methods: polynomial-based cubature, pseudo-Monte Carlo (PMC), and quasi-Monte Carlo (QMC). The QMC method uses non-random, uniformly distributed sequences, leading to faster convergence and lower integration error. The Halton sequence is used in the QMC method due to its simplicity and effectiveness. The MMNL model is generalized from the multinomial logit model by integrating over the distribution of unobserved random parameters. It allows for flexible substitution patterns and accommodates unobserved heterogeneity in individual sensitivity to exogenous variables. The paper compares the performance of the three methods using numerical experiments. Results show that the QMC method, particularly with the Halton sequence, provides more accurate estimates with significantly fewer draws and computational time compared to the PMC and cubature methods. The QMC method is especially effective in higher dimensions, making it a valuable tool for estimating the MMNL model in practice. The study concludes that the QMC method is more efficient and practical for estimating the MMNL model, suggesting its broader application in discrete choice modeling.This paper proposes using a quasi-random sequence for estimating the mixed multinomial logit (MMNL) model. The MMNL model is a flexible discrete choice model that accommodates general patterns of competitiveness and heterogeneity in individual sensitivity to exogenous variables. Traditional estimation methods, such as pseudo-random maximum simulated likelihood, evaluate multi-dimensional integrals in the log-likelihood function by averaging integrand values at pseudo-random points. The authors suggest an alternative quasi-random maximum simulated likelihood method using uniformly distributed sequences, which provides better accuracy with fewer draws and computational time. Numerical experiments on intercity travel mode choice show that the quasi-random method outperforms the pseudo-random method. This has potential to significantly influence the practical use of the MMNL model, especially given its flexibility in accommodating complex behavioral structures. The paper discusses three numerical integration methods: polynomial-based cubature, pseudo-Monte Carlo (PMC), and quasi-Monte Carlo (QMC). The QMC method uses non-random, uniformly distributed sequences, leading to faster convergence and lower integration error. The Halton sequence is used in the QMC method due to its simplicity and effectiveness. The MMNL model is generalized from the multinomial logit model by integrating over the distribution of unobserved random parameters. It allows for flexible substitution patterns and accommodates unobserved heterogeneity in individual sensitivity to exogenous variables. The paper compares the performance of the three methods using numerical experiments. Results show that the QMC method, particularly with the Halton sequence, provides more accurate estimates with significantly fewer draws and computational time compared to the PMC and cubature methods. The QMC method is especially effective in higher dimensions, making it a valuable tool for estimating the MMNL model in practice. The study concludes that the QMC method is more efficient and practical for estimating the MMNL model, suggesting its broader application in discrete choice modeling.