It is difficult to estimate accurately the cost of constructing a large scientific instrument that involves many techniques. On the other hand, most of the component parts of the VLBA consist of antennas and electronic systems that already exist or are being fabricated. The kind of 25-m antennas being constructed for the VLA will cost about $900,000 each and will work at wavelengths as short as 1 cm. A multifrequency radiometer, hydrogen maser frequency standard, small control computer, control building, and wide-band instrumentation recorder bring the cost to about $1.5 million per element, or $15 million for a ten-element array using tape recorders. A multistation playback facility, with ten recorders and enough correlators to handle all interferometer pairs simultaneously, together with the necessary computers to control the processor and reduce the data, may add $5 million. The total cost is thus about $20 million at current prices, including an adequate supply of magnetic tape. This is comparable to the cost of existing large radio telescopes and arrays. An array that used a geostationary communication satellite to transmit the data to a real-time correlator would cost $30 million to $50 million more, but this is still within the price range of other space astronomy projects. It is thus feasible to construct at reasonable cost an intercontinental very long baseline array which has sub-milliarcsecond resolution. This would complement the Very Large Array now being constructed (4), which is much more sensitive to objects of low surface brightness. This next step would permit the study of the universe with unprecedented angular resolution.