On July 2, 1937 (midnight GMT) Earhart and Noonan took off from Lae in the heavily loaded Electra. Their intended destination was Howland Island, a flat sliver of land 6,500 ft (2,000 m) long and 1,600 ft (500 m) wide, 10 feet (3 m) high and 2,556 miles (4,113 km) away. Their last known position report was near the Nukumanu Islands, about 800 miles (1,300 km) into the flight. The United States Coast Guard cutter Itasca was on station at Howland, assigned to communicate with Earhart's Lockheed Electra 10E and guide them to the island once they arrived in the vicinity.
Final approach to Howland Island
Through a series of misunderstandings or errors (the details of which are still controversial), the final approach to Howland Island using radio navigation was not successful. Fred Noonan had earlier written about problems affecting the accuracy of radio direction finding in navigation. Some sources have noted Earhart's apparent lack of understanding of her Bendix direction-finding loop antenna, which at the time was very new technology. Another cited cause of possible confusion was that the USCG cutter Itasca and Earhart planned their communication schedule using time systems set a half hour apart (with Earhart using Greenwich Civil Time (GCT) and the Itasca under a Naval time zone designation system).
Motion picture evidence from Lae suggests that an antenna mounted underneath the fuselage may have been torn off from the fuel-heavy Electra during taxi or takeoff from Lae's turf runway, though no antenna was reported found at Lae. Don Dwiggins, in his biography of Paul Mantz (who assisted Earhart and Noonan in their flight planning), noted that the aviators had cut off their long-wire antenna, due to the annoyance of having to crank it back into the aircraft after each use.
During Earhart and Noonan's approach to Howland Island the Itasca received strong and clear voice transmissions from Earhart identifying as KHAQQ but she apparently was unable to hear voice transmissions from the ship. At 7:42 a.m. Earhart radioed "We must be on you, but cannot see you—but gas is running low. Have been unable to reach you by radio. We are flying at 1,000 feet." Her 7:58 a.m. transmission said she couldn't hear the Itasca and asked them to send voice signals so she could try to take a radio bearing (this transmission was reported by the Itasca as the loudest possible signal, indicating Earhart and Noonan were in the immediate area). They couldn't send voice at the frequency she asked for, so Morse code signals were sent instead. Earhart acknowledged receiving these but said she was unable to determine their direction.
In her last known transmission at 8:43 a.m. Earhart broadcast "We are on the line 157 337. We will repeat this message. We will repeat this on 6210 kilocycles. Wait." However, a few moments later she was back on the same frequency (3105 kHz) with a transmission which was logged as a "questionable": "We are running on line north and south." Earhart's transmissions seemed to indicate she and Noonan believed they had reached Howland's charted position, which was incorrect by about five nautical miles (10 km). The Itasca used her oil-fired boilers to generate smoke for a period of time but the fliers apparently did not see it. The many scattered clouds in the area around Howland Island have also been cited as a problem: their dark shadows on the ocean surface may have been almost indistinguishable from the island's subdued and very flat profile.
Whether any post-loss radio signals were received from Earhart and Noonan remains controversial. If transmissions were received from the Electra, most if not all were weak and hopelessly garbled. Earhart's voice transmissions to Howland were on 3105 kHz, a frequency restricted to aviation use in the United States by the FCC. This frequency was not thought to be fit for broadcasts over great distances. When Earhart was at cruising altitude and midway between Lae and Howland (over 1,000 miles (1,600 km) from each) neither station heard her scheduled transmission at 0815 GCT. Moreover, the 50-watt transmitter used by Earhart was attached to a less-than-optimum-length V-type antenna. 
The last voice transmission received on Howland Island from Earhart indicated she and Noonan were flying along a line of position (taken from a "sun line" running on 157–337 degrees) which Noonan would have calculated and drawn on a chart as passing through Howland. After all contact was lost with Howland Island, attempts were made to reach the flyers with both voice and Morse code transmissions. Operators across the Pacific and the United States may have heard signals from the downed Electra but these were unintelligible or weak.
Some of these transmissions were hoaxes but others were deemed authentic. Bearings taken by Pan American Airways stations suggested signals originating from several locations, including Gardner Island. It was noted at the time that if these signals were from Earhart and Noonan, they must have been on land with the aircraft since water would have otherwise shorted out the Electra's electrical system. Sporadic signals were reported for four or five days after the disappearance but none yielded any understandable information. The captain of the USS Colorado later said "There was no doubt many stations were calling the Earhart plane on the plane's frequency, some by voice and others by signals. All of these added to the confusion and doubtfulness of the authenticity of the reports."