At present, many inversion methods are available for ideal geometrical models, but few data interpretation approaches can be directly used for the horizontal pipeline of finite length. In this paper, a forward modeling method is proposed to calculate the gravity gradient tensor of the horizontal pipeline. The theory is to deduce the forward formulas of a horizontal pipeline of finite length from the existing formulas of a horizontal cylinder of infinite length. The calculated tensors at different measuring heights are analyzed in shapes and numerical values. In addition, we introduce two boundary extraction functions—HGA and V zz -HDR to identify the boundary edges of the horizontal pipeline. Due to the high resolution and high sensitivity of gravity gradient tensor data, analytic signal has the advantage of locating anomalous bodies. In the case of the fixed measuring distance d, varied half extending length L, and outside diameters R, forward contour maps of the analytic signal of pipeline tensor, HGA and V zz -HDR are analyzed for boundary recognition. The result shows a good accuracy in length recognition in all cases and a varied effect in outside diameter recognition by the ratio of d and R. Euler deconvolution of the analytic signal of gravity gradient tensor is a geophysical inversion method, which can estimate the source location automatically or semi-automatically without a priori information, and we succeed in applying this inversion method to the pipeline model. Considering the effect of the noise, the convergence of inversion result is related to L and R, but the source location is always located on the geometric center with good accuracy.