We implement the proof of principle for the quantum walk of one ion in a linear ion trap. With a single-step fidelity exceeding 0.99, we perform three steps of an asymmetric walk on the line. We clearly reveal the differences to its classical counterpart if we allow the walker/ion to take all classical paths simultaneously. Quantum interferences enforce asymmetric, non-classical distributions in the highly entangled degrees of freedom (of coin and position states). We theoretically study and experimentally observe the limitation in the number of steps of our approach, that is imposed by motional squeezing. We propose an altered protocol based on methods of impulsive steps to overcome these restrictions, in principal allowing to scale the quantum walk to several hundreds of steps.PACS numbers: 03.67. Ac, 05.40Fb, 0504Jc A quantum walk[1] is the deterministic quantum mechanical extension of a classical random walk. A simple classical version requires two basic operations: Tossing the coin (coin-operation), allowing for two possible and random outcomes. Dependent on this outcome, the walker performs a step to the right or left (stepoperation). In the quantum mechanical extension the operations allow for coherent superpositions of entangled coin and position states. After several iterations the probability to be in a certain position is determined by quantum mechanical interference of the walker state that leads to fundamentally different characteristics of the walk [2].The motivation for studying quantum walks is twofold. On the one hand, many classical algorithms include random walks. Examples can be found in biology, psychology, economics and physics, for example Einstein's simple model for Brownian motion [3]. The extension of the walk to quantum mechanics might allow for substantial speedup of related quantum versions [2], as in prominent algorithms suggested by Shor[4] and Grover [5] due to other quantum-subroutines. On the other hand, the quantum walk could lead to new insights into entanglement and decoherence in mesoscopic systems [6]. These topics might be explored by increasing the amount of walkers -even before any algorithm might benefit from the quantum random walk.Quantum walks have been thoroughly investigated theoretically and first attempts at implementation have been performed with a very limited amount of steps due to a lack of operation fidelity or fundamental restrictions within the protocol. Some aspects have been realized on the longitudinal modes of a linear optical resonator [7] and in a nuclear magnetic resonance experiment [8]. An implementation based on neutral atoms in a spin-dependent optical lattice[9, 10, 11] has resulted in an experiment recently. Other proposals considered an array of microtraps illuminated by a set of microlenses [12] and Bose-Einstein condensates [13]. Travaglione and Milburn[14] proposed a scheme for trapped ions to transfer the high operational fidelities [6] obtained in quantum information processing (QIP) into To implement the deterministic "tossing of t...