please look here:

http://humanresearch.at/newwebcontent/?page_id=221&lang=en

you can record 3 days without problems and with one battery. Our longest recording lasted 9 days, uninterrupted.

This looks interesting, impressive sampling rate. Thank you.

BodyGuards are capable of this I believe.

http://www.heartmath.org/templates/ihm/e-newsletter/publication/2011/spring/hrv-recording-device.php

there are dozens of monitors using Holter ECG approach. Check nevrocard.eu for their products (I bought from them BRS software, but I know they are very good at others also) or CIMVA software. They are device-independent and let you determine a plethora of HRV measures.

The actiheart can record for long periods of time depending on which parameters you are interested in. http://www.camntech.com/products/actiheart/actiheart-overview

Suunto Memory belt is able of saving up to 1 million beats (approx. 6 days), can still be bought, but is no longer produced. As RR-detection should be the same as in the t6 accuracy should be sufficient (https://www.researchgate.net/publication/41911339_Comparison_of_three_mobile_devices_for_measuring_R-R_intervals_and_heart_rate_variability_Polar_S810i_Suunto_t6_and_an_ambulatory_ECG_system)

sorry if I insert here my suggestion.

I am not here to outline you the importance of the HRV current studies and I have not to remember you that the current linear or non linear methodologies are not completely satisfactory. Of course every day we find papers published also in valuable international journals that continue to use standard linear indexes and FFT or DFT for their clinical investigations and conclusions. Non linear methodologies have their valuable content also if currently they still do not find a continuous clinical appreciation. Usually three bands are considered in the frequency domain , as you well know, the VLF, the LF , and the HF for short 5-6 minutes recording. We need to remain in this framework, we need to use such bands for our estimations but my modest opinion is that the basic methodology as well as the physiological setting need to be advanced. This is the reason because we have studied for several years a new method, called the CZF method. We have verified, however, that such method, although of great interest, results of difficult interpretation for some clinicians. Consequently we have elaborated a new method, the new CZF version , that , in addition to the rigour of the previous formulation , gives results that give immediate clinical interpretation.

The method as well the software for your experimentation are given in our site www.saistmp.com Consequently I invite all you to visit the site and go to use such new method in your experimental as well as clinical conditions and verify directly its interest. The present software is the primary elaboration. Within a brief time we will give the complete version responding to all the formulation of the method. All the researchers that , at this time, will result using currently such new CZF method will receive obviously automatically the updates.

A final question. We have experienced the method on a large number of subjects and in some different pathological conditions. Consequently don’t hesitate to contact me if you need help, suggestions, questions, previous results that may help you in your use of the method ([email protected])

I verify that my previous answer was not added. We may furnish R-R recording device at a very low cost and obviously all the necessary siftware inclluding those for analysis of variability as explained in our site www.saistmp.com. In case contact [email protected]

I agree with previous comments, the first thing is to decide if you want raw data, ECG, or R-R intervals that is easier to store. ECG will depend on sample frequency that increases a lot the file. Anyway, I think is not complicated to download every day the device for a few minutes. It doesn't seems to compromisse the collection of data. But this dependes on the protocol.