Physical and mathematical phonation models commonly rely on a quasi-one-dimensional flow model. The assumption of quasi-one-dimensional flow through a glottis with fixed length is analyzed for different cross-section shapes: Circle, rectangle, ellipse, and circular segment. A simplified flow model is formulated which accounts for kinetic losses, viscosity, and cross-section shape. It is seen that the cross-section shape cannot be neglected since it alters boundary layer development and hence the viscous contribution to the pressure drop across the glottis. The commonly applied quasi-one-dimensional flow model is shown to be inaccurate, indicating the potential benefit of taking into account the cross-section shape.