Real-time 2D ultrasound systems are used routinely in every hospital and are a huge success both technically and commercially. This paper discusses the signal processing problems that needs to be tackled in order to move from 2D to 3D real-time ultrasound systems. The first problem discussed is that of handling 2000 10000 elements in the transducer. Sparse array methods is a way to reduce the number of elements and cost without compromising quality. Examples of performance with sparse arrays are presented. The second important problem is that of frame-rate. In 3D the frame-rate will be so low that real-time acquisition will be impossible unless some form of parallelism is exploited. Various ways of doing that such as multiple receive beams, coded transmit excitation and limited diffraction beams are discussed.