On October 1, 2020, at 2004 eastern daylight time, a Beechcraft M35 Bonanza, N656V, was substantially damaged when it was involved in an accident in Green Sea, South Carolina. The private pilot and the passenger sustained minor injuries. The airplane was operated as a Title 14 Code of Federal Regulations Part 91 personal flight. The pilot stated that, while in cruise flight at 5,500 ft mean sea level as the airplane approached the intended destination, he listened to the airport terminal information service (ATIS) recording, moved the fuel selector from the auxiliary tank position to the right tank position, and then requested a descent from air traffic control, which was granted. The pilot stated that when he disconnected the autopilot to initiate the descent, the engine “went back to idle, like someone had pulled the throttle back.” He said that he moved the propeller, throttle, and mixture levers full forward, checked switch positions, and tried all positions on the fuel selector, and power was not restored. He “bumped” the fuel boost pump to the ON position then turned it off to prevent flooding of the engine. The pilot informed the controller of the engine power loss, and the controller advised the pilot of the nearest airport, which was Green Sea Airport (S79). The airplane collided with trees and terrain about 1 mile west of S79, resulting in substantial damage. The airplane was recovered to an aircraft recovery facility, where it was examined by Federal Aviation Administration (FAA) inspectors and the engine was removed for a test run at a commercial engine repair facility. Continuity of the fuel lines and the fuel selector were confirmed using compressed air while rotating the fuel selector through each position. The air flowed freely without obstruction. After removal, the engine was placed in a test stand, where it started immediately, accelerated smoothly, and ran continuously without interruption. The mechanic who conducted the test run under FAA supervision provided a brief written statement describing what tasks were performed, as well as the results that were witnessed when they started and ran the engine. The mechanic was then contacted by telephone to clarify some of the language in his statement. He explained that the engine was rpm-limited (2,200 rpm) because the fuel mixture was too rich, and it was not possible to determine which part of the fuel system required adjustment to reach full operating rpm, or whether the differences were in the rigging of the test stand vs the airplane, or due to impact damage. The JPI engine data monitor (EDM) was removed, and examination of the downloaded data revealed temperatures and fuel flow rates consistent with those described by the pilot. Automatic dependent surveillance – broadcast (ADS-B) data could not be correlated with the data from the EDM, so alignment of the data was estimated at key points along the flight profile (Level-off, cruise altitude, final descent) with changes in fuel flow consistent with those profile changes. At level off, and during the cruise segment about 5,100 ft pressure altitude, the airplane showed a steady fuel flow rate of about 13 gallons per hour (gph). About 6 minutes and 30 seconds before the end of the EDM data plots, the fuel flow decreased from 13.1 gph in 6-second intervals to 12.9, 11.2, 0.2, and 1.6 gph before fuel flow reached zero. Fuel flow remained at zero for about 1 minute, when an individual 0.2-gph data plot displayed. Fuel flow returned to zero gph for about 1 minute 30 seconds and then displayed fuel flow rates of 0.1, 0.4, 0.4, 0.8, 0.3 over a 30-second span before it returned to zero for the remainder of the flight. A review of the fuel system with the airplane manufacturer revealed that the fuel flow transducer was downstream of the fuel selector and prior to the flow divider of the fuel injection system, where it measured metered fuel. With fuel supplied through the fuel selector, the engine-driven fuel pump (with a windmilling propeller) or an operating electric fuel boost, with the throttle open, and the mixture set anywhere but at “cutoff,” would provide measurable fuel flow and pressure. Removal and inspection of the fuel selector was performed under the supervision of an NTSB investigator. Compressed air was used to test return lines and supply lines through each fuel selector setting (OFF, LH TANK, AUX TANK, RH TANK). The air flowed freely with no obstructions noted. Each position displayed a positive detent. With the fuel selector placed in the OFF position, or any position between detents, the fuel selector was fully obstructed and would not flow air. A placard on the accident airplane’s fuel selector plate stated, “Warning: Position Selector in Detents Only. No Fuel Flow to Engine Between Detents.”