This paper introduces Trajectory Consistency Distillation (TCD), an improved method for latent consistency distillation that enhances image quality and detail in text-to-image synthesis. TCD consists of two key components: the *trajectory consistency function* (TCF) and *strategic stochastic sampling* (SSS). The TCF broadens the scope of the self-consistency boundary condition by using trajectory mapping and an exponential integrator, reducing parameterization and distillation errors. SSS explicitly controls stochasticity and suppresses accumulated discretization errors, improving sampling quality. Experiments show that TCD significantly enhances image quality at low function evaluations (NFEs) and outperforms the teacher model at high NFEs, demonstrating superior performance in both visual quality and detail. The method is versatile and can be applied to various models, including conditional models and community models, with minimal changes.This paper introduces Trajectory Consistency Distillation (TCD), an improved method for latent consistency distillation that enhances image quality and detail in text-to-image synthesis. TCD consists of two key components: the *trajectory consistency function* (TCF) and *strategic stochastic sampling* (SSS). The TCF broadens the scope of the self-consistency boundary condition by using trajectory mapping and an exponential integrator, reducing parameterization and distillation errors. SSS explicitly controls stochasticity and suppresses accumulated discretization errors, improving sampling quality. Experiments show that TCD significantly enhances image quality at low function evaluations (NFEs) and outperforms the teacher model at high NFEs, demonstrating superior performance in both visual quality and detail. The method is versatile and can be applied to various models, including conditional models and community models, with minimal changes.