In this work, we introduce GaussianCut, a new method for interactive multiview segmentation of scenes represented as 3D Gaussians. Our approach allows for isolating the desired object by interacting with a single view. It accepts intuitive user input such as clicking on the object, coarse scribbles, or text. We choose 3D Gaussian Splatting as the underlying representation as its explicit nature makes it easier to extract a subset of Gaussians that encode the object(s) of interest. GaussianCut models the Gaussians representing a scene as a graph and proposes an energy function on the graph that can be minimized effectively using graph cut. Our proposed energy function combines the user inputs with the inherent properties of the scene and uses graph cut to partition the Gaussians as foreground and background. In order to obtain an initial coarse segmentation, we leverage the advancements made in 2D image and video segmentation models and further refine these estimates using our graph construction. Our empirical evaluations show the adaptability of GaussianCut across a diverse set of scenes. It achieves competitive performance with state-of-the-art approaches for 3D segmentation without requiring any additional segmentation-aware training and effectively handles foreground elements of varying shapes and sizes.