Curarized and non‐curarized rat hemidiaphragm muscles were indirectly stimulated in vitro. 2. The fluid bathing the active curarized muscles was eluted through a dextran gel (Sephadex G‐10), effecting a complete separation of ACh from curare. The acetylcholine fraction was then assayed on an isometric leech muscle preparation. 3. Prostaglandin (PGE1) in a concentration fifteen times that estimated to be released from the skeletal muscle preparation did not affect the response of leech muscle to ACh. 4. The amount of ACh released by curarized muscles (4‐9 X 10(‐18) mole/impulse‐junction) was not significantly different from that released by non‐curarized muscles (4‐6 X 10(‐18) mole/impulse‐junction). These quantities are similar to those obtained by previous workers. It is concluded that curare in a paralytic dose does not affect the output of ACh from motor nerve terminals stimulated at low frequencies. 5. Spontaneous release of ACh from non‐curarized muscles was estimated at 0‐45‐0‐65 p‐mole/min. hemidiaphragm. It is calculated that only 2% of this amount could give rise to post‐synaptic electrical events, the remainder having a non‐synaptic source. 6. The number of molecules of 'quantal' ACh released by stimulated muscle is calculated as 2‐5 X 10(6)/impulse‐junction, taking account of the non‐synaptic release. The number of ACh molecules in one quantum was estimated to be 6250, an amount that could be easily accommodated in one synaptic vesicle.