Compressive sensing is well known for its robust signal reconstruction ability from a smaller set of samples than required according to Nyquist criterion. In this paper compressive sensing (CS) has been proposed in eigenspace for Multichannel Electrocardiogram (MECG) signals. Principal component analysis (PCA) is used to give eigenspace signals. PCA functions twofold here: First it confines the diagnostic information of MECG signals spread over different channels to few eigenspace signals, and furthermore it gives sparser signals to be explored further. The sparsity of the eigenspace MECG signals (N samples), is further enhanced by representing them in orthogonal wavelet basis. CS is then used to collect few random measurements (M samples, M < N) of these sparse signals using a random sensing matrix with independent identically distributed (i.i.d.) entries taken from sampling a Gaussian distribution. The signal recovery from these few measurements has been carried out by a convex optimization problem using L1-norm minimization. The quality of reconstruction of the recovered signal has been found satisfactory. Performance of the proposed algorithm has been evaluated in terms of percentage root mean square difference (PRD), normalized root mean square difference (NRMSD), normalized maximum amplitude error (NMAX), and maximum absolute error (MAE). Lowest PRD value, 4.61% has been obtained for lead V5 after simulation using CSE multi-lead measurement library database.