“PIC16F72A” arrives in enquiries more often than it appears on a datasheet. Microchip’s current catalogue does not carry a device under that exact name. What does exist is three different parts that get called that, and which one you actually have changes what the job actually is.

On an OTP part there is exactly one chance. The image can be recovered but the original cannot be rewritten, so a recovered file is verified by programming a fresh 16F72 and running the board — there is no way to test on the source device.
On the flash part you can re-program the same chip, compare behaviour before and after, and confirm the image is complete by running it rather than by trusting a checksum alone.
On the 16F722A the instruction set is still the mid-range 35-instruction core, but the register map and configuration words differ from the 16F72. A recovered image cannot simply be dropped onto the other device. It can still be disassembled and understood, and that is usually what the customer needs when the goal is to maintain the product rather than to duplicate it.
The top marking is the first evidence: Microchip’s OTP parts carry a C in the part number, the flash parts an F, and date codes bracket the changeover. The second is the device identifier, which the programming interface normally reports even when code protection is set — protection stops the memory from being read, not the device from identifying itself.
If neither is conclusive, the board itself usually settles it. A design with a socketed 28-pin part and a 1990s date code is an OTP board; one with a soldered SOIC and a 2005 date code is not.
The package is opened, the array is probed directly, and the image is reconstructed and cross-checked against a second read. That process is identical across all three devices — what changes is the verification route and what you can do with the result. The file itself is delivered in the formats described under firmware extraction. Die-level work is described under IC decapsulation; if what you need is a working replacement rather than a file, IC cloning is the better starting point.
A photograph of the marking, the board reference and any documentation you have, even if it is a user manual. A manual tells us what the product is supposed to do, which is often the fastest way to confirm that a recovered image is complete rather than merely readable.
The PIC16F72 page covers the flash part in its own right, including why the absence of data EEPROM matters when you rebuild a board. For the 40-pin members of the same generation, see the PIC16F77.
We work for customers who own the hardware or hold authority to repair it, under a non-disclosed agreement. If the part turns out to be one we cannot help with, we will say so before any work is charged.