someone asked "A product with P89C51xxx with IAP, works under similar conditions and face the similar power up sequence, but I have not come accross inadvertent flash corruption." The main reason is that those devices have a narrow operating voltage 4.5 to 5.5V, as I recall. There is a brownout detect circuit with a trip point close to the lower Vdd limit. As Vdd decreases this circuit hold the part in reset. Typically it will hold it to a relatively low Vdd level, say 2V. If it stops holding the part in reset, there typically isn't enough Vdd th charge-pump up to 12V to do the programming so its a non issue. Meanwhile the 5V charge that was across the reset cap is discharging (of course you have that reverse-biased diode in your reset circuit to discharge the cap when the power is removed, don't you?). The reset input on the 5V devices are active high and have a high VIH, about 0.7Vdd, or about 3.5V. So if you dropped the system voltage down to 2V or so and powered back up, you would get a valid reset (the voltage on the reset pin will follow Vdd as long as Vdd is changing). Also the brownout detect will be working until a voltage that is higher than the reset VIH.